UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Some roles for expert systems in planning Colby, Lisa J. 1990

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-UBC_1990_A8_2 C64.pdf [ 7.25MB ]
Metadata
JSON: 831-1.0098478.json
JSON-LD: 831-1.0098478-ld.json
RDF/XML (Pretty): 831-1.0098478-rdf.xml
RDF/JSON: 831-1.0098478-rdf.json
Turtle: 831-1.0098478-turtle.txt
N-Triples: 831-1.0098478-rdf-ntriples.txt
Original Record: 831-1.0098478-source.json
Full Text
831-1.0098478-fulltext.txt
Citation
831-1.0098478.ris

Full Text

S O M E ROLES FOR EXPERT SYSTEMS IN PLANNING by LISA J. COLBY A THESIS SUBMITTED IN T H E REQUIREMENTS MASTER PARTIAL FULFILMENT O F FOR T H E D E G R E E OF O F ARTS in T H E F A C U L T Y OF G R A D U A T E STUDIES School of Community and Regional Planning We accept this thesis as conforming to the required standard T H E UNIVERSITY O F BRITISH COLUMBIA September 1990 Lisa J. Colby, 1990 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at The University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the Head of my Department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. School of Community and Regional Planning The University of British Columbia 2075 Wesbrook Place Vancouver, Canada V6T 1W5 Date: September 1990 ABSTRACT This thesis explores whether computer-based expert systems can be used in planning and, if so, under what circumstances. Expert systems are computer-based programs that solve problems in a way that mimics the human reasoning process. Expert system reasoning relies upon logic and rules-of-thumb rather than the numerical and mathematical algorithms of most other computer programs. Planning has been interpreted in the professional context of urban and regional planning, rather than the cross-disciplinary fiscal or project management planning often implied in computer literature. To determine expert systems could be useful to planners, the reasons for incompatibility between the nature of planning and conventional computers models of the past are explored. Advantages and disadvantages of expert systems are considered. Expert systems represent substantial improvement in areas where conventional programs are inadequate. Chapters 2 to 5 form the theoretical base of the thesis. Chapter 2 explains the fundamentals of expert system reasoning and how it differs from other computer software. Chapter 3 outlines reasons why this technology might appeal to planners. Chapter 4 introduces some of the disadvantages of expert systems, including technical limitations, ethical and legal issues. Chapter 5 introduces general guidelines to help the reader understand what type of planning tasks might benefit from the use of this new tool. Chapters 6 to 8 consider issues raised in the preceding chapters. Three Canadian ii systems now at the forefront of expert systems applications to planning are reviewed in chapters 6, 7 and 8 Respectively. The first application, HERMES, is an emergency planning application. It advises emergency response personnel during crises involving hazardous materials. SCREENER is an environmental planning application. It assists environmental officers at Transport Canada to assess simple capital projects for environmental impact statements and screen out more complicated ones for further review. The third application used for illustrative purposes, PLANCHECKER, is a municipal planning example. The system assists plan checkers at City Hall in assessing building plans. The three case-studies satisfy the task suitability guidelines quite well and appear to be successful applications of expert systems to planning. It is still too early to draw definite conclusions, but it is likely the technology will prove useful to planners. Planners should be prepared understand both the potential and limitations of expert systems so they can use the technology wisely. iii ACKNOWLEDGEMENTS I would like to acknowledge and thank certain individuals who helped in the development of this thesis. My supervisor, Dr. Henry Hightower offered valuble and welcome guidance throughout. Thanks also to Dr. Craig Davis who offered a second perspective and constructive criticism. I am grateful to the private consultants, Dr. Edwin Bluett, Dr. Eric Heikkila, and Mr. Tim Webb, who gave freely of their time to explain and demonstrate their programs. I am also grateful to the many other professionals listed in the appendix who volunteered responses to my research questions. Finally, thanks to my family for their moral support, patience, and encouragement iv T A B L E OF CONTENTS ABSTRACT ii A C K N O W L E D G E M E N T S iv Chapter 1. INTRODUCTION 1 1.1. RATIONALE 1 1.2. M E T H O D O L O G Y 2 1.2.1. Literature Review 2 1.2.2. Direct Contact 3 1.3. SCOPE A N D LIMITATIONS 4 1.4. THESIS ORGANIZATION 5 Chapter 2. W H A T A R E EXPERT SYSTEMS? 7 2.1. INTRODUCTION 7 2.2. DESCRIPTION 7 2.2.1. Knowledge Base 9 2.2.1.1. Knowledge Acquisition 10 2.2.1.2. Knowledge Representation 12 2.2.2. Inference Engine 13 2.2.3. User Interface 15 2.2.4. Working Memory 15 2.2.5. Additional Sources of Information 16 2.3. DIFFERENCES BETWEEN EXPERT SYSTEMS A N D CONVENTIONAL PROGRAMS 16 2.3.1. Separate Knowledge Base and Inference Engine 16 2.3.2. Inferencing 16 2.3.3. Programming Language 17 2.3.4. Uncertainty and Incomplete Information 19 2.3.5. Accountability 20 2.4. M U L T I - P R O G R A M POTENTIAL 21 2.5. CHAPTER S U M M A R Y 21 Chapter 3. W H Y SHOULD EXPERT SYSTEMS APPEAL TO PLANNERS? 23 3.1. INTRODUCTION 23 3.2. PLANNERS' LOSS O F FAITH IN COMPUTERS AS DECISION AIDS 23 3.3. A MISMATCH O F TOOLS TO TASK 24 3.4. SEMI-STRUCTURED PLANNING TASKS A N D INFORMAL INFORMATION 25 3.5. T H E FIT O F EXPERT SYSTEMS TO PLANNING THEORY ... 27 3.5.1. Utopianism 28 3.5.2. Rationalism 28 3.5.3. Incrementalism 29 3.5.4. Method Planning 30 3.6. EXPERT SYSTEMS OPEN NEW DOORS 31 3.6.1. Main Advantages of Expert Systems for Planners 32 3.6.2. Additional Advantages of Expert Systems 33 3.7. CHAPTER S U M M A R Y 37 v Chapter 4. WHY EXPERT SYSTEMS MIGHT NOT APPEAL TO SOME PLANNERS 41 4.1. INTRODUCTION 41 4.2. SPECIFIC LIMITATIONS O F EXPERT SYSTEMS 41 4.3. G E N E R A L CONCERNS REGARDING EXPERT SYSTEM USE 43 4.3.1. Legal 44 4.3.2. Ethical Issues 45 4.3.3. Human Resistance to Change 45 4.3.4. Appropriate Context 46 4.4. CHAPTER SUMMARY 46 Chapter 5. IDENTIFYING TASKS IN PLANNING WHICH W O U L D SUIT EXPERT SYSTEM T E C H N O L O G Y 49 5.1. INTRODUCTION 49 5.2. GUIDELINES FOR DETERMINING TASK SWT ABILITY TO EXPERT SYSTEMS 49 5.2.1. Discussion 52 5.3. EXPERT SYSTEM PROBLEM TYPES 55 5.3.1. DIAGNOSIS A N D REPAIR 56 5.3.2. DESIGN A N D PLANNING 56 5.3.3. INTERPRETATION 57 5.3.4. MONITORING A N D C O N T R O L 57 5.3.5. INSTRUCTION 58 5.4. PLANNING APPLICATIONS 58 5.5. CHAPTER SUMMARY 61 Chapter 6. HERMES ILLUSTRATIVE APPLICATION 64 6.1. HERMES - ORIENTATION 64 6.1.1. Developers 64 6.1.2. Function 64 6.1.3. Users 65 6.1.4. Scope 66 6.1.5. Technical 66 6.2. PROBLEM SUITABILITY 67 6.2.1. Characteristics 67 6.2.2. Suitability Criteria 68 6.3. PERFORMANCE 76 6.3.1. Main Features 76 6.3.1.1. User Interface 76 6.3.1.2. SCENE Window 76 6.3.1.3. SUGGESTED THINGS TO D O window 77 6.3.1.4. ADVISED E M E R G E N C Y ACTION GIVEN window 77 6.3.1.5. W A T C H THESE LEVELS window 77 6.3.1.6. Tutorial system 78 6.3.2. Satisfaction 78 6.3.3. Future 79 6.3.3.1. Knowledge base 79 6.3.3.2. Uncertainty 80 vi 6.3.3.3. Environmental Models 80 6.3.3.4. Sense of time 81 6.3.3.5. Explanation facility 81 6.4. CHAPTER SUMMARY 82 Chapter 7. SCREENER ILLUSTRATIVE APPLICATION 86 7.1. SCREENER - ORIENTATION 86 7.1.1. Developers 86 7.1.2. Function 86 7.1.3. Users 87 7.1.4. Scope 87 7.1.5. Technical 89 7.2. PROBLEM SUITABILITY 89 7.2.1. Problem Characteristics 89 7.2.2. Suitability Criteria 90 7.3. PERFORMANCE 99 7.3.1. Features 99 7.3.2. Satisfaction 100 7.3.3. Future 100 7.4. CHAPTER SUMMARY 101 Chapter 8. P L A N C H E C K E R ILLUSTRATIVE APPLICATION 104 8.1. P L A N C H E C K E R - ORIENTATION 104 8.1.1. Developers 104 8.1.2. Function 104 8.1.3. Users 105 8.1.4. Scope 106 8.1.5. Technical 106 8.2. PROBLEM SUITABILITY 107 8.2.1. Problem Characteristics 107 8.2.1.1. Description 107 8.2.1.2. Difficulties Associated with the Work 108 8.2.2. Suitability Criteria 109 8.3. PERFORMANCE 116 8.3.1. Main Features 116 8.3.2. Satisfaction 116 8.3.3. F U T U R E 117 8.4. CHAPTER SUMMARY 117 Chapter 9. CONCLUSION 119 9.1. SUMMARY 119 9.1.1. Objective 119 9.1.2. Description 119 9.1.3. Advantages 120 9.1.4. Disadvantages 120 9.1.5. Suitability Guidelines 121 9.1.6. Hermes 123 9.1.7. Screener 124 9.1.8. Planchecker 126 9.2. C O N C L U D I N G C O M M E N T 128 vii REFERENCES 131 APPENDIX A - Sample Letters and Lists of Professionals Contacted 138 APPENDIX B Sample Screen from HERMES 151 APPENDIX C Sample Reports from SCREENER 152 viii CHAPTER 1. INTRODUCTION 1.1. RATIONALE The main purpose of this research project is to determine the characteristics of a successful relationship between planning work and computer-based Expert Systems. More specifically, what type of planning tasks are suitable to the use of expert systems technology and what types are not? The term planning is used in the context of community and regional, environmental, or emergency planning, rather than cross-disciplinary project management, spreadsheet, or fiscal planning that the word often implies in computer literature. Expert systems are the most recent form of computer technology to be applied to planning. They mimic human problem-solving by their use of non-numeric, heuristic knowledge. Most planners know little, if anything, about expert systems and are thus in a poor position to make decisions about how, when or whether to use this technology in their work. While expert systems research tended to focus on fields such as medicine and chemistry prior to the 1980's, there has been much diversification since then. The field of planning has received considerable attention recently. Several prototypes are now under development and some commercially available planning applications should appear soon. Planners may wish to be knowledgeable about current expert systems technology so that they can use it wisely. Planners also have a role to play in the shaping and development of these systems if the product is to be a genuinely useful planning tool. 1 INTRODUCTION / 2 This thesis is intended to raise practitioners' understanding of how expert systems technology might fit into their field. 1.2. M E T H O D O L O G Y The methodology adopted for the development of this thesis included the following stages: * Familiarization with expert systems. * Familiarization with the nature of decision-making in planning. * Developing an understanding of what makes certain tasks more suitable to expert systems use than others. * Search for examples of planning related expert systems. * Selection and review of three applications for illustration purposes. * Observations from illustrative applications. Information was obtained from the literature and by direct contact with individuals involved in expert system research and development 1.2.1. Literature Review The literature reviewed included articles, journals, books, and program documentation. The review was intended to provide a better understanding of the nature of expert systems as well as the issues of concern as they related to planning. Most materials were obtained from local research libraries. Materials were also borrowed from individuals encountered in the research process and through an interlibrary loan network. INTRODUCTION / 3 1.2.2. Direct Contact Direct contact was made with many researchers and planners to try to unearth recent developments in the field which might not yet have been reported in journals and books. The Planning Educators Electronic Mail Network directory (PLANET) was used to contact a wide range of planning academics. Letters were sent to all those listed except those few who were known to be uninvolved with computers. Letters were also sent to the directors of all Canadian university Planning Departments to probe the state of planning related expert system research in Canadian universities (Appendix A). The literature review uncovered the names of some expert system authors, as well as planners otherwise involved in computer technology. Those individuals were contacted by mail or phone and were asked for their programs, documentation, reports, or suggestions as to where more expert systems might exist, on disk or paper. Several software companies were contacted (Appendix A) in the hope that they had developed and started to market planning applications. Catalogues and price lists including any commercially available planning applications were requested. Samples of letters sent to the different groups can be found in Appendix A The criteria for selecting illustrative applications were: • the availability of disks and documentation and the system authors' responses to questions, • the relevance to planning, • the degree to which the application illustrated a different planning emphasis than other examples chosen. While several prototypes in planning are being developed, the availability of disks and documentation proved to be the limiting factor. Very few systems of significance are operable or readily available. INTRODUCTION / 4 1.3. SCOPE AND LIMITATIONS This research was done on a limited budget and within a limited time frame which undoubtedly affected the number of contacts made in the survey and prevented the acquisition of certain materials. For reasons of both time and money, therefore, the literature review formed a large part of the basis of the project Those people surveyed were selected contacts, not a representative sample of all planners using expert systems. Some names may have been missed by the author's literature review and contact network. Several of those contacted did not respond. There may also be a number of planning expert systems in operation or well along in the development phase which are not yet reported in the literature. This became evident from survey results and word-of-mouth accounts of expert systems activity which did not appear in the literature review. The survey cannot be considered random because not all planning expert systems were known and considered. It is beyond the scope of this thesis to provide a comprehensive survey of planning related expert systems activity occurring in Canada or around the world. As one could explain the potential uses and suitability of a telephone in planning without teaching planners how to build one, this thesis attempts to explain the potential and suitability of expert systems. The text should be comprehensible to practitioners and students with minimal computer knowledge. It is beyond the scope of this thesis, and the qualifications of the author, to provide INTRODUCTION / 5 a graduate level critique of expert systems from a technical, computer science perspective. The perspective remains that of a planning student and potential expert system user. 1.4. THESIS ORGANIZATION Chapter 1 explains the rationale, methodology, scope, limitations, and the organization of this thesis. Chapter 2 explores the nature of expert systems. It defines them and explains how they work. Chapter 3 suggests how expert systems can be appropriate to planning and why they should appeal to planners. The advantages and potential influence of such technology on planning philosophy are explored. Chapter 4 outlines some of the concerns associated with expert systems and what planners should not expect from them. Chapter 5 identifies specific areas of planning which could benefit from expert systems technology. Guidelines for evaluating a task's suitability to expert system use are suggested and a problem typology is outlined. Examples of a range of planning applications are provided. Chapters 6, 7 and 8 each cover a particular example of a planning related expert system. A description of each sample application is given. Each application is discussed INTRODUCTION / 6 in light of issues raised in preceding chapters and their task characteristics are considered in the context of the suitability criteria from Chapter 5.0. Chapter 9 summarizes the thesis and finishes with commentary on the prognosis for expert system use in the planning field. CHAPTER 2. WHAT ARE EXPERT SYSTEMS? 2.1. INTRODUCTION This chapter will describe expert systems to the user. A fundamental grasp of the nature of the technology should help the reader understand the advantages, disadvantages and potential applications of the technology as discussed later in the thesis. Expert systems are first described in a very general way. Each of the basic components are then discussed separately to give the user a better understanding of how they work together. Those features which distinguish expert systems from previous technology are then explained. 2.2. DESCRIPTION An expert system is a computer based program which aims to solve a given problem in much the same way as would a human expert (Waterman, 1985). This represents a different approach to problem solving than that used by other types of computer programs, (collectively referred to as 'conventional' computer programs throughout this thesis). Expert systems are comprised of a base of knowledge on the one hand including rules, facts and expertise on a given subject, and an inferencing mechanism on the other hand to work logically with these rules. Expert systems offer advice for solving problems and making decisions, based on the expertise of human experts (Ortolano and Perman, 1987). They are applied to subject areas in which problems would be difficult to solve by non-experts. The user would enter all available information of a given problem as prompted by the system. Initial possibilities according to the knowledge base are determined by the system and 7 WHAT A R E EXPERT SYSTEMS? / 8 confirmation or negation of these possibilities function to narrow in on the solution. Conceptual models within the system suggest tests to be performed or queries to direct to the user. The range of possible solutions is gradually whittled down to a conclusion which is valid according to the system's knowledge base. Classical theories of logical deduction developed by mathematicians and philosophers play a significant role in this process which essentially relies upon a combination of rules and facts to draw a conclusion (Ortolano et al., 1987). Many problem areas are beyond the reach of conventional computer programs because they use heuristic reasoning. Heuristic reasoning means reasoning based on handy strategies, rules of thumb, judgments and intuition which cannot be translated into mathematical algorithms. Mathematical algorithms are the stepby-step numerical procedures which conventional computers follow to reach a correct answer. Expert systems are better equipped to handle problems involving heuristics because of their ability to represent decision making symbolically rather than mathematically. Human experts do not generally solve problems through sets of equations or laborious mathematical computations. Instead, they represent problem concepts symbolically and then use various strategies and heuristics to work with these concepts (Waterman, 1985). It would be untrue to suggest that experts do not use mathematical procedures, but the emphasis in problem solving tends to be upon the manipulation of symbols. An expert may sidetrack to perform a calculation but the result of that mathematical process can be used as a symbolic concept This concept may then be manipulated along with other concepts according to heuristic rules or strategies learned through WHAT A R E EXPERT SYSTEMS? / 9 experience. Such reasoning may not lead to an optimal solution but will often find a useful solution quickly (Ortolano et al., 1987). Expert systems also represent knowledge symbolically. They do this by using a symbol (a string of characters) to represent a concept like, house, rezoning applicant, or a particular piece of numerical data. Combinations of these symbols (called 'symbol structures') can describe real world situations and relationships. In order to solve problems, the expert system manipulates these symbols rather than carrying out mathematical computations (Waterman, 1985). An expert system has three basic components: * a knowledge base * an inference engine (control mechanism) * a user interface The knowledge base is a reservoir of domain expertise. The inference engine is the logical controlling mechanism which allows the system to work with the knowledge base. The user interface is the medium which allows the computer and the person using it to communicate with each other. It allows users to express themselves in graphics or English-type prose, for example. A working memory and additional sources of information are also commonly included components. Each of the system components will now be discussed in detail. 2.2.1. Knowledge Base The knowledge base is at the heart of an expert system (Mick and Wallace, 1986). It embodies the expertise of a given domain and, like a database, is usually stored on a disk-file. It is a collection of facts, rules, definitions and computational procedures. The W H A T A R E EXPERT SYSTEMS? / 10 knowledge is ultimately both explicit and accessible, unlike most conventional programs (Waterman, 1985). Expertise is explicitly declared in the form of rules in the knowledge base, and the priority and strategies for using those rules is also explicitly declared in the inferencing component of the program. The accessibility of the knowledge is made possible through the explanation feature. Expert systems can explain to users (upon request) how they arrived at a conclusion. In the interest of system validity, the proper acquisition and representation of knowledge become crucial issues. Effective elicitation of the necessary knowledge and decision process, and an accurate representation of that knowledge in the system are necessary if the results are to be reliable and useful. 2.2.1.1. Knowledge Acquisition Knowledge can be acquired from many sources including books, reports, databases, and empirical data, but the primary and dominant source is usually a human expert, On the subject of consulting human reasoning and expertise for building the knowledge base, the following can be said (Waterman, 1985, p.5): "An expert is a person who because of training and experience is able to do things the rest of us cannot; experts are not only proficient but also smooth and efficient in the actions they take. Experts know a great many things and have tricks and caveats for applying what they know to problems and tasks; they are also good at plowing through irrelevant information in order to get at basic issues, and they are good at recognizing problems they face as instances of types with which they are familiar. Underlying the behaviour of experts is the body of operative knowledge we have termed expertise. It is reasonable to suppose, therefore, W H A T A R E EXPERT SYSTEMS? / 11 that experts are the ones to ask when we wish to represent the expertise that makes their behaviour possible." (Paul E. Johnson, Scientist) In addition to vast domain specific factual knowledge, human experts use heuristic reasoning, their own rules of thumb, to shortcut lengthy arguments, resulting in greater efficiency. Choosing the right concepts and respective relationships is what allows human experts to process complex questions in their heads without suffering memory overload and unduly impairing problem-solving ability (Newton and Taylor, 1986). Experts often rely on many patterns of knowledge to guide them through a problem. Their mental shortcuts may be the result of complex reasoning based on significant data and experience. They can efficiently work through complicated situations without re-examining each step in the reasoning process. It is the ability to recognize information that is basic, relevant and not requiring re-examination which makes them experts (Waterman, 1985). Developers of expert systems attempt to replicate the heuristic and scientific knowledge characteristic of human experts by extracting and encoding their procedures, strategies and problem solving knowledge. Unfortunately, human expert knowledge can be difficult to interpret Human experts are often incapable of articulating each step of their own reasoning process. The paradox of expertise refers to the observation that, "...the more competent experts become, the less able they are to describe the knowledge they use to solve problems (Waterman, 1985, p.154)." Studies have also shown that when pressed for an explanation, experts WHAT A R E EXPERT SYSTEMS? / 12 will invent, "... plausible lines of reasoning that bear little resemblance to their actual problem solving activity (Waterman, 1985, p.154)." A common technique to help experts declare their knowledge involves systematic interviews in which the knowledge engineer (program developer) 'talks' the expert through problem solving situations. An incomplete, rudimentary skeleton of the knowledge base can be established. Experts' reactions to successive runs of the program for different sample problems can then further direct and focus the knowledge base as rules are added and changed. The knowledge base and thus the expert system prototype itself can be improved incrementally with each successive run. This evolutionary process, sometimes referred to as prototyping, has been found to be the most effective way to build an expert system (Waterman, 1985). 2.2.1.2. Knowledge Representation The most effective way of representing knowledge is still the subject of research but a common and very effective method is through declarative rules. Each rule is treated as an independent statement of knowledge, proving only the truth of its consequent (Newton and Taylor, 1988). When the consequent represents a problem solution it is known as a goal. In a large knowledge base where the consequents of many rules are only intermediate steps, they are known as sub-goals which then fit into other rules, chaining rules into a tree (Newton, Taylor and Sharpe, 1988). Two examples of rules in a knowledge base would be the following (Clark, Chang and Sidebottom, 1988, p.19): WHAT A R E EXPERT SYSTEMS? / 13 Example 1. TJF substance is liquid AND substance is not flammable AND fire type is high intensity or impinging T H E N set evacuation distance to 300m. Example 2. IF substance is gas AND substance is not flammable, not corrosive, or not poisonous AND firetype is low intensity or high intensity T H E N set evacuation distance to 800m. The simplest forms of rules allow only complete propositions joined by the word AND. More sophisticated forms allow the use of other logical operators such as NOT, OR; the use of English words like is and includes; and the inclusion of numerical variables with conventional conditional operators like equals, less than, and greater than (Newton et al., 1988). This type of representation allows for, a cross-linking between rules. 2.2.2. Inference Engine In order to effectively mimic the performance of a human expert in a particular area, knowledge must be represented in a way which allows the system to reason (Newton et al., 1986). The inference engine, also known as the controlling mechanism, fulfills this function. It is a set of procedures for manipulating the knowledge base information in order to reach conclusions. This includes procedures for determining which rules and procedures in the knowledge base to examine first and which facts to seek from the user (Ortolano et al., 1987). The inference engine is separate from the knowledge base. The inference engine either draws all possible conclusions out of the knowledge base, or tries to prove that a given statement can be inferred from the knowledge base. W H A T A R E EXPERT SYSTEMS? / 14 The following example illustrates how inferences can be made: If it is known that A is true and also that the rule If A then B holds, then we may infer that the conclusion B holds true. B might be a condition in other rules which will lead to further conclusions, and so on (example from Newton et al.. 1988, p.267). Some forms of knowledge are about other knowledge, such as what rules apply to a certain question, or in what order given rules should be tried to solve a problem efficiently. Expert systems can use either forward chaining or backward chaining (or both). In instances of forward chaining, data is entered and the system is asked to devise a conclusion or suggest a procedure (Clark et al., 1988). For example, forward chaining is used to generate advice in the Heuristic Emergency Response Management System (HERMES) case study, Chapter 6. Backward chaining starts with a conclusion and the system is then asked to determine the missing information or provide reasons why such an outcome has occurred (Clark et al., 1988). By working backwards from T H E N to IF, unsatisfied conditions determine the missing information and the user is given suggestions as to how to fill in the missing details (Clark et al., 1988). The separation of the inference engine or control mechanism from the knowledge base is the basis for the generic software known as expert system shells. A shell consists of the general control mechanism and an editing facility for entering the detailed WHAT A R E EXPERT SYSTEMS? / 15 knowledge base of a given domain (Ortolano et al., 1987). This allows people who do not want to become involved in the details of the inference engine to develop an expert system by simply providing the knowledge base. Different shells of varying levels of sophistication are commercially available. 2.2.3. User Interface A person would interact with an expert system in much the same way as s(he) would interact with a human expert whom s(he) had chosen to consult for the same problem. A consultation would take place between the user and the expert system in which the user explains the problem, and the system queries the user until it narrows in upon a recommended solution. The user interface facilitates this exchange of information by translating human language to machine language and vice-versa. End-users (people using the final system) can express themselves much better through graphics, menus and the typed word than through machine language. They can also better understand the computer's response when it is translated into this form. 2.2.4. Working Memory The term working memory is used in the context of software, as opposed to hardware memory storage, and refers to a temporary storage feature. It is distinct from the knowledge base and contains only information generated during the current run of the system (Ortolano et al., 1987). This might include information from conventional numerical programs, remote sensors, data bases, or input from the user. WHAT A R E EXPERT SYSTEMS? / 16 2.2.5. Additional Sources of Information Some expert systems provide linkages so they can include facts and information from sources other than their own knowledge bases e.g. conventional numerical analysis programs, data bases, or remote sensors. These additional sources often require special interfaces for use with the system. 2.3. DIFFERENCES BETWEEN EXPERT SYSTEMS AND CONVENTIONAL PROGRAMS While some people would argue that expert systems are just another type of computer program (Miller and Walker, 1988), there are certain features which distinguish them as a class unto themselves. 2.3.1. Separate Knowledge Base and Inference Engine An expert system's knowledge base is encoded and kept separate from the inference engine (underlying control structure). This fact differentiates expert systems from conventional computer programs. It allows the user to modify the knowledge base without rewriting substantial portions of the program as would be necessary in traditional computer modeling. Modification would simply involve adding, removing or changing a rule. The inferencing procedures and other rules would thus remain intact This feature also facilitates the prototyping process of developing expert systems. 2.3.2. Inferencing The use of inferencing rather than mathematical algorithms is perhaps the most distinctive characteristic of expert systems (Newton, 1986). The 'reasoning' of conventional computer programs consists of sequentially processed statements. Decision WHAT A R E EXPERT SYSTEMS? / 17 and jump statements may interfere with this flow somewhat but basically each statement is followed by the next (Newton et al., 1988). This is not the case with expert systems. In expert systems, principles of symbolic logic are used to draw conclusions from facts about the premises of various rules (Kim, (Wiggins and Wright, 1990). The processing of one statement may result in a question to the user, the answer to which may be put through a whole number of statements in a chain reaction. A search routine would then determine which is the next best statement to process, in the interest of efficiently arriving at a valid conclusion to the user's problem (Newton et al., 1988). 2.3.3. Programming Language It has been argued that expert systems are just another programming language, albeit one which can handle non-numeric programming (Rodriguez-Bachiller, 1990). It has also been argued that the programming language used to implement the expert system is not a defining characteristic (Ortolano et al., 1987; Kennedy 1987; Waterman 1985). The confusion may be a result of the term programming language being used in slightly different ways. It should be noted that expert systems have been developed in several different programming languages. It is the perception of this author that those declaring expert systems as simply another programming language, are in actual fact challenging the possibly sensational claim that expert systems are intelligent, rather than disputing that expert systems can be differentiated from other types of programming. Their point may simply be that expert systems are not really intelligent, but merely a type of programming which WHAT A R E EXPERT SYSTEMS? / 18 incorporates a kind of reasoning not accommodated by other computer software. While expert systems do represent a different type of programming language, they cannot always be defined by the specific language in which they are programmed. Expert systems organize their knowledge in a declarative rather than a procedural form and this predisposes them toward certain languages more often than others. Kennedy suggests that there are two types of computer languages, both of which can be used to develop expert systems, although one is generally more suitable (1987). The first type is algorithmic, or procedural, and the second type is symbol-manipulative and declarative. Expert systems are usually developed in declarative rather than procedural languages. Languages like FORTRAN, COBOL, PL/1, Pascal are algorithmic. They are problem oriented languages, meaning they were designed for particular classes of problems. FORTRAN, for example, is tailored for performing algebraic calculations and is highly suited to scientific, mathematical and statistical problem areas (Waterman, 1985). P R O L O G and LISP are declarative, logical languages and process symbols (character strings) efficiently. These languages, also known as 'symbol manipulation languages' were designed for Artificial Intelligence applications (Waterman, 1985) Planners have often been frustrated that computers and their major languages were algorithmic, in the sense that everything must be couched in a step by step, procedural manner (Kennedy, 1987). Problems such as whether or not a particular WHAT A R E EXPERT SYSTEMS? / 19 action is permitted according to a set of regulations, are not very well suited to an algorithmic approach. Laws and regulations are more declarative in nature even though they may describe a procedural, algorithmic process (Kennedy, 1987). Another frustration of planners is that many of the more conventional languages are closer to mathematics than natural (e.g. English) or graphical languages. Declarative, symbol-manipulative languages allow planners to work in a 'quasi-English' rather than 'quasi-mathematical' notation (Kennedy, 1987). Most expert systems work is done in LISP or PROLOG because of their flexibility at symbol manipulation, but it is possible to develop one in a problem oriented language such as FORTRAN, Pascal, etc. (Waterman, 1985). 2.3.4. Uncertainty and Incomplete Information Most rule-based systems, such as expert systems try to work in spite of uncertain and incomplete information. Through inexact reasoning, an expert system could focus upon and solve a problem, which conventional programs would find virtually impossible. It could produce results of "human-level ability" (Mick et al., 1986). Methods used include fuzzy logic developed by L Zadeh, and other ad-hoc procedures (Kim et al., 1990). Fuzzy logic is an approach to inexact reasoning whereby the inference rules are approximate, allowing better manipulation of imprecise or incomplete information; and measures of confidence are rated in approximate linguistic terms. Truth values such as true, very true, not very true, or many and few are used (Waterman, 1985). WHAT A R E EXPERT SYSTEMS? / 20 The measure of confidence can also be a numeric weight which begins as a measure of confidence in the input data. It is then propagated through the system to measure the strength of rules being applied. Unfortunately, the strength of rules is usually an arbitrary assignment given by human experts during the system design, and is therefore subject to error (Mick et al., 1986). The particular arithmetic process for calculating probabilities throughout a system is still the subject of debate (Newton et al., 1988). Other methods for carrying uncertainty through the system have also been used. Some reflect instinct but have little statistical basis. Expert system uncertainty features are still underdeveloped, and research is ongoing in this area. 2.3.5. Accountability Yet another distinctive feature of expert systems is their transparency or accountability. This refers to the ability to explain reasoning in terms understandable to the user (Ortolano et al., 1987). During the consultation, a user may want to know why or how a particular conclusion was reached, or why a particular question is being asked. The expert system may be able to display the chain of reasoning behind its actions in an English type format. In addition to providing clarification to the user, the advantage of the accountability feature, is that the user may have more confidence in the system's judgement The rules and hence the values and assumptions, are not as difficult for the user to access as they might be in a conventional program. WHAT A R E EXPERT SYSTEMS? / 21 2.4. M U L T I - P R O G R A M POTENTIAL An expert system can be a collection of programs and software, including both the problem solving component and the support component, that work together to solve a problem in a specific domain. The support environment may perform countless functions such as linking the user to the main program, providing debugging tools for the system building process, or providing sophisticated graphic facilities to improve the user-interface (Waterman, 1985). For this reason, it is an expert system rather than an expert program. 2.5. CHAPTER SUMMARY Expert systems are computer-based software systems designed to replicate the decision-making process of human experts. They can incorporate heuristic reasoning. The essential feature of heuristic reasoning is the use of 'rules-of-thumb,' judgments, experience and hunches. An expert system is basically comprised of a knowledge base, an inference engine and some form of user-interface. The knowledge base is a collection of rules, definitions and procedures in a given field of expertise. The inference engine is the reasoning mechanism which drives the system. It manipulates and uses the knowledge base, determining which rules to call upon and when. The user interface is simply the communication medium between the user and the computer itself. The more sophisticated and 'friendly' it is, the simpler it is for an inexperienced user to make use of the system. WHAT A R E EXPERT SYSTEMS? / 22 Expert systems work in an interactive consultative manner much as would a human expert They cannot work alone because they continually need information and judgments from the user. The main features which distinguish expert systems from the large body of other computer software are: the knowledge base and 'reasoning' are kept separate; the inferencing method does not depend upon numerical, mathematical algorithms; expert systems can use inexact reasoning and work with incomplete information; expert systems are 'transparent' allowing user access to their reasoning both for questions asked and advice given. The language used to develop the expert system is not its defining characteristic. Expert systems can be programmed in various languages, but declarative languages such as LISP or PROLOG are most commonly used. These languages developed specifically for artificial intelligence use, manipulate symbols more efficiently than conventional algorithmic languages such as F O R T R A N and Pascal. Arguments in the literature which suggest that expert systems are 'just another programming language,' may be raising another issue, simply challenging the presumed claim of having created truly 'intelligent' systems. While expert systems may be closer in reasoning style to human experts than conventional programs, it may be an exaggeration to suggest that they possess true intelligence. CHAPTER 3. WHY SHOULD EXPERT SYSTEMS APPEAL TO PLANNERS? 3.1. INTRODUCTION This chapter explains why expert systems might appeal to planners. Reasons for planners' initial disappointments with computer aided decision making models are suggested. The notion that, in many cases, conventional programs were ill-suited to their applications is discussed. The importance of informal information in planning is considered, and expert systems are then discussed in the context of different planning paradigms. A list of six principal reasons why planners should find expert systems appealing as work tools, followed by sixteen general advantages, is then provided. 3.2. PLANNERS' LOSS OF FAITH EN COMPUTERS AS DECISION AIDS While many conventional computer models and programs remain suited to certain clearly structured problem areas, many planners have lost faith in computers as tools for decision-making (Langendorf, 1985). The loss of faith may be largely a perception of broken promises. The early excitement surrounding the use of computers in planning in the 60's and early 70's focussed on the development of large scale land-use and transportation models. Due to consistently poor results, many of these models were considered disappointing failures. By 1973 when the American Planning Association published an article entitled "Requiem for Large-Scale Models," the loss of enthusiasm was widespread throughout the profession. It has been suggested that this loss of faith may have been due to a mismatch of tools to task. Conventional models were ill-suited to many planning tasks and, had this been recognized, many disappointments could have been avoided. This subject is 23 WHY SHOULD EXPERT SYSTEMS APPEAL TO PLANNERS? / 24 worth discussing in more detail. 3.3. A MISMATCH OF TOOLS TO TASK The literature offers many reasons to explain why planners lost faith in computers and why conventional models were often inappropriate as decision making tools. These are outlined in Table 1 (Langendorf, 1985). Table 1. REASONS PLANNERS LOST FAITH IN CONVENTIONAL COMPUTER MODELS • Decision makers do not understand models. (The models' reasoning processes are hidden and inaccessible to the user). • Decision makers often cannot specify in advance what they want They require trial and error and a sequential decision-making process that conventional models typically do not accommodate. • Decision-making involves change and models often lack the flexibility to respond to fluctuating needs. • Decision making often involves judgmental and other "soft" criteria, multiple criteria or objectives, and individual or group preferences that the formal models typically do not accommodate. • Decision makers had the perception that models assume unrealistic quantitative precision. Planners apparently perceived quite correctly that the tools were inappropriate to planning decision-making realities. The computer models reflected a scientific, rational, decision-making bias. While this coincided with the bias of planners in the 60's, perceived problems of reliance on the purely 'rational' planning paradigm were surfacing. The models were being discovered, as was the purely rational paradigm, to be incapable of handling many planning situations which involved judgment and changing circumstances. Several studies since then have pointed out the much more complex nature of WHY SHOULD EXPERT SYSTEMS APPEAL TO PLANNERS? / 25 decision-making in the field of planning. The theory of incrementalism received increased recognition from planners. The value and role of heuristics and informal information was introduced. 3.4. SEMI-STRUCTURED PLANNING TASKS AND INFORMAL INFORMATION Information is valued in planning because it can reduce the inherent uncertainties in problems and thereby facilitate decision-making (Hopkins & Schaeffer in, Han and Kim, 1989). It has been suggested that data and information form the very basis of the profession's activities because they describe the conditions of the real world (Harris, 1987 in Han et al., 1989). The practitioner must work with this information to try to achieve the goals and aspirations of the client (Han et al., 1989). Public practitioners need information to continually advise and assist the public in defining and achieving their aspirations. Information is collected, processed and analyzed in order to understand the complex environment where the planning activities are to take place. Presumably, the more this environment is understood, the better are the opportunities for effective planning and decision-making (Han et al., 1989). Many planning problems require judgment, preferences and have no possible single optimal solution, because of various constraints and multiple objectives. These are sometimes referred to as unstructured or semi-structured problems (Langendorf, 1985). Information and knowledge garnered from non-formalized procedures such as hearsay, intuition, professional hunches or personal experiences (informal information) can be important in planning decisions (Han et al., 1989). WHY SHOULD EXPERT SYSTEMS APPEAL TO PLANNERS? / 26 Other planning problems are the opposite. They require no judgement and there is a possible optimal solution. These are sometimes referred to as structured problems (Langendorf, 1985). It is crucial to recognize that most computer models used by planners until now have been designed for structured problems, and yet most decision-making in planning is of a semi-structured or unstructured nature (Langendorf, 1985). Most urban information systems until now, however, have excluded the use of this more subjective informal information. While this may be largely due to values contributed by the rational and scientific school of thought, it is also due in part to the technical limitations of conventional computerized information systems (Han et al., 1989). It is the structured problems that have been the easiest to automate with conventional computer technology. It is the semistructured and relatively unstructured problem areas which expert systems can help planners to address. They are tools designed to work with the benefit of human experts' judgement, experience and rules of thumb. Expert systems can be integrated with conventional systems, augmenting rather than replacing them in many cases. Database management systems (DBMS), Geographic Information Systems (GIS), and Decision Support Systems (DSS) are familiar forms of computerized urban information systems (UIS). The use of subjective, non-formalized knowledge which has hitherto been excluded from conventional UIS might now be taken advantage of to complement formal knowledge (Han et al., 1989). WHY SHOULD EXPERT SYSTEMS APPEAL TO PLANNERS? / 27 3.5. T H E FIT OF EXPERT SYSTEMS T O PLANNING THEORY It seems to be generally agreed upon that there is no single planning theory espoused by all planners (Catanese, 1988). There has been much debate and experimentation in planning from utopianism, to rational thought, incrementalism to methodism planning. It is perhaps unlikely that there ever will be consensus on a single philosophy, particularly given the breadth of planning activity and the many roles the planner may be asked to fulfill. Despite severe criticism of the profession's earlier reliance upon the rational school of thought, for example, such an approach can be effective in certain circumstances. Many of the different planning theories may be useful in the appropriate context and the adoption of one theory to the exclusion of all others may leave a planner ill-equipped to deal with the range of tasks at hand (Catanese, 1988). The planner's best strategy may be to try to understand the benefits and drawbacks of each planning approach, and to develop the skill to choose an appropriate combination for a given situation. Likewise, the planner must understand the nature and limitations of any tools used in the planning process, as they may exert significant influence on process results. The unsuccessful experiences of trying to use computer models in situations to which they were not suited underscores the necessity of knowing when a tool is appropriate to a task. Expert systems, as a tool are worth discussing in this context WHY SHOULD EXPERT SYSTEMS APPEAL TO PLANNERS? / 28 3.5.1. Utopianism In terms of utopianism, there is little likelihood that expert systems can be of much use. Expert systems have little to offer when problems require imaginative and creative visions to inspire others toward a goal, or when complete intellectual leaps into uncharted territory are required. They can be no more creative then the people who designed them and the processes they employ are merely those already conceived of and declared by the human experts. 3.5.2. Rationalism The true rationalist may not appreciate expert systems' incorporation of heuristic reasoning into the decision making process. To those who recognise certain limitations of the 'rational approach,' however, expert systems can be a favourable complement The rational approach refers to the planning paradigm in which a system and its problems are analyzed, all possible alternative solutions are identified, the likely results of each are evaluated, and the best one is selected. This paradigm has been at the root of most conventional planning computer models and programs. The philosophy's basic assumption that all relevant information about a situation can be discovered and analyzed, and that all options can be conceived of and appropriately evaluated before the correct decision must be taken, is still its principal weakness (Catanese, 1988). In those few situations, where most conditions are fairly well known, a few simple goals have been established, and the means for accomplishing them are clearly understood, a truly rational approach to problem solving may work. In such cases, conventional computer programs can be perfectly acceptable and may be a more WHY SHOULD EXPERT SYSTEMS APPEAL T O PLANNERS? / 29 ideologically compatible work tool. The ideological bent of expert systems, it must be understood, is to be able to incorporate judgmental and subjective factors into decision-making. They offer increased and more realistic human problem solving potential to rational components, when incorporated into the automation process. Conventional algorithmic models can be incorporated into expert systems wherever desired, but the rule-based system as a whole also acknowledges the fact that information may be missing or uncertain. Expert systems offer reasoning to account for their decisions and, acknowledging that there may not be a best solution, concentrate on Finding acceptable solutions, given the goals and constraints. 3.5.3. Incrementalism As work tools, expert systems may be advantageous to the incrementalist approach to planning. Often referred to as the science of muddling through, the specific means to an end may not be known even though many may agree on the ultimate goal. The existence of a knowledge base separate from the inference engine allows for relatively easy modification of the system. This could allow for a continuously evolving decision making process as planners grapple with a problem, in which rules are added, changed or removed from the knowledge base. The W H A T IF feature of expert systems also may allow a user to incrementally approach a problem solution, trying different input alternatives and combinations. Many problems of this type may be unlikely to be encountered often enough to justify the time and money necessary for expert system development Others may. Ideologically speaking, the potential of incremental adjustments to the system as values WHY SHOULD EXPERT SYSTEMS APPEAL T O PLANNERS? / 30 and circumstances change, is an important one. There is not the built-in assumption that circumstances and solution techniques are cast in stone and the common problem of computer models becoming obsolete is reduced. The incremental approach and expert systems also share the recognition that all information is not necessarily available, complete, or even certain. The use of a tool which acknowledges such issues even to a degree, relieves the burden on the system user somewhat, to constantly account and haphazardly try to adjust whatever results may be produced. Expert systems can be designed to deliver probability estimates with any results it provides. The accuracy and reliability of such results are vulnerable to criticism but there is an attempt to accommodate the uncertainty factor. Perhaps the most significant aspect of expert systems is their propensity to satisfice. Rather than finding an optimal solution to problems which often have no undeniably correct or best solution, the expert system attempts to provide a reliable method of generating satisfactory advice under the circumstances. As a tool this certainly reflects the reality of decision- making in many semi-structured or unstructured planning tasks. 3.5.4. Method Planning This approach refers to planning activities for which the method is clear but the specific ends to be achieved are often undefined and unknown (Catanese, 1988). These activities are often tedious, repetitious and vulnerable to sloppy execution by bored or harried personnel. The procedure itself, however, allows for the generation of information that might have otherwise never been considered. Zoning reviews, public hearings, building code appeals, and annual surveys are examples of such tasks. W H Y SHOULD EXPERT SYSTEMS APPEAL TO PLANNERS? / 31 While the somewhat ritualistic rule-based aspect of method planning is well suited to expert systems, the information seeking emphasis of these activities by nature, is not They seek information from people which computers alone could not give. There are tasks within the larger approach which could benefit from expert systems, though. Expert systems tend to want to narrow in upon a specific problem and solution. Method planning may be able to benefit from expert systems if the problem domain became the procedure itself. Problems such as how to proceed under certain circumstances, or in the face of changing events, for example, could be ideal applications which would not have been possible with conventional urban information systems. 3.6. EXPERT SYSTEMS OPEN NEW DOORS In view of the reasons for which planners rejected conventional large scale models (Table 1) and the nature of planning decisionmaking, Langendorf suggested a number of characteristics which would extend the range of problems that computers could reliably address. These characteristics are outlined in Table 2. Table 2. CHARACTERISTICS WHICH WOULD EXPAND COMPUTER USE IN PLANNING • an iterative and interactive process of defining problems, goals, and methods of achieving a solution. • the ability to handle multiple objectives. • the ability to incorporate preferences. • the ability to incorporate judgments. • the flexibility to adapt to changing conditions. • the ability to work with incomplete information. • the ability to deal with uncertainty. Interestingly enough, expert systems share all the characteristics of this 'wish list'. The degree to which they reflect each characteristic varies, but they are clearly better suited to many planning problems than were their conventional model predecessors. WHY SHOULD EXPERT SYSTEMS APPEAL T O PLANNERS? / 32 These attributes of expert systems alone have been sufficient to attract considerable attention in planning recently. 3.6.1. Main Advantages of Expert Systems for Planners This section itemizes the advantages which expert systems can offer to planners. Six principal features which may appeal to planners are summarized in Table 3. They are interesting in the context of Langendorfs explanation of planners' disappointment with computers, the necessary characteristics for expanding the range of computers in planning, and the nature of decision-making in planning. Table 3. MAIN REASONS EXPERT SYSTEMS SHOULD APPEAL TO PLANNERS 1. Expert systems can expand the range of planning tasks which could be automated. Tasks which involve heuristics or informal information may now be suitable for automation. (It should be recognized that this may not be considered an advantage by some practitioners). 2. The reasoning processes of expert systems are accessible. The user can query the system for explanation of its process, ensuring greater user confidence. 3. Expert systems can support an iterative and interactive process of problem resolution. They can accommodate trial and error decision-making. 4. Expert systems are flexible enough to be adapted to changing conditions. This is possible through the separate and relatively easily modified knowledge base. Rules can be added or removed without damaging the reasoning of the system or the knowledge base. This eliminates the problem of the system becoming obsolete during development or shortly thereafter. By-laws can change, for example, even during the course of building an expert system. The flexibility of this technology means that by changing that particular rule or adding a new one to reflect the new by-law, the system is usable when it is finished. 5. Expert systems can incorporate "soft" judgmental and preference information in decision-making. WHY SHOULD EXPERT SYSTEMS APPEAL TO PLANNERS? / 33 This is supported through heuristic rules, and the ability to ask the user for judgments directly. 6. Expert systems do not assume unrealistic quantitative precision. They attempt to work with incomplete and uncertain information and try to reflect realistic confidence values for advice givea 3.6.2. Additional Advantages of Expert Systems There are additional reasons why expert systems may appeal to planners. The advantages (7-22) outlined below are commonly cited benefits of expert systems (Miller et al., 1988; Newton et al., 1988; and Waterman, 1985). They are not specific to planning and can be important to any profession or area of expertise. 7. Expert systems can preserve specialists' knowledge. Organizations can lose specialist knowledge when experts change jobs, retire, or do not use their knowledge regularly. Rarely-used expertise such as emergency response skills may be forgotten because the human mind must use its expertise or things will be forgotten. Expert systems do not forget, and their knowledge remains when in-house human expertise leaves the organization. The expert system can be built upon by subsequent expert personnel, and its accumulated knowledge base can serve as an 'institutional memory.' 8. Expert systems are ideal for training purposes (Newton et al., 1988). Staff turnover and the introduction of new procedures, by-laws or regulations, can result in constandy high training costs and poor quality work. An expert system can mitigate these problems if used as a tutor or as an expert assistant for personnel learning the ropes. Its problem solving and explanatory features can help the user understand how, why and according to what policies decisions are taken. Novices can not unwittingly skip key steps. WHY SHOULD EXPERT SYSTEMS APPEAL TO PLANNERS? / 34 9. Unnecessary burden is relieved on experts, supporting them in the more creative aspects of their work (Miller et al., 1988). Experts are often distracted from their real work because too many enquiries are directed at them. An expert system could handle the straightforward enquiries, freeing the human expert for more problematic situations which only they are qualified to tackle. This advantage may have to be weighed against the distraction of experts from their real work if they are spending time on the development of an expert system. Systems can take as long as five years to build. There are trade-offs to be considered when evaluating whether or not to invest in the development of an expert system. 1 0 . Expert systems are not bewildered by growing amounts of data. As the volume of information on a given subject grows, the computer can easily manipulate and integrate amounts of data that would overwhelm and bewilder the human expert Through various interfaces, an expert system can coordinate several sources of data in addition to its own knowledge base. 1 1 . Expert systems impartially investigate all hypotheses because they have no personal prejudices except those specifically declared in the knowledge base. While expert systems can take advantage of human instinct and judgement in problem solving, they are not susceptible to prejudiced execution of the process. They will not bend the problem or advice given to fit a pre-established hunch about the solution. There are cases where the user may challenge the solution offered by the computer based on a hunch, and wish to override its advice, but the computer still provides a good check for the user in this sense. The user may wish to think again. 1 2 . Interaction with an expert system can take place at a user's own conceptual WHY SHOULD EXPERT SYSTEMS APPEAL TO PLANNERS? / 35 level, by virtue of the system's deep domain understanding. Users who are proficient in the problem domain may use the system more as an expert colleague on with whom to check ideas. They may not require explanations on most steps of the process. Or they may only need an explanation to a certain level. Naive users may require explanation more consistently and to more than one level of depth. From this point of view, the expert system doesn't waste the more proficient user's time, and doesn't push the novice through procedures which they don't understand. 13. Expert systems are consistent in their reasoning and decision making (Waterman, 1985). Various factors such as stress, distractions, haste, or emotional changes can cause a person to handle identical problems differently. Expert systems are not subject to these problems and the user of the system can feel confident that their problem was put through the same full procedure as other problems of its type. There can be no blaming the system for having been in a bad mood, sloppy or lazy. 14. Expert systems are a productivity multiplier, making specialist knowledge available to others, even when the specialist is busy (Newton et al., 1988). More people including less experienced employees could do the work which used to be done by those with specialized knowledge. From the perspective of a builder trying to get plans approved in a busy planning department, this would be good news. To employees who depend upon a plentiful workload for their jobs, or good employees whose unusual speed and accuracy earns them higher status and pay, this feature may be perceived as a disadvantage. They may lose advantages they previously enjoyed. Expert system development and implementation should try to resolve conflicts such as these. Again, trade-offs are involved. WHY SHOULD EXPERT SYSTEMS APPEAL T O PLANNERS? / 36 15. Expert systems are more efficient than other automation techniques because of their ability to determine what aspects of the current problem are critical and to devote most of the computational resources to solving them (Miller et al., 1988). Only those rules which are necessary for the problem at hand are invoked and the computer does not waste time querying the user on other issues. 16. The reasoning of an expert system is easy to document (Waterman, 1985). The reasoning of an expert system can be explained in 'quasi-English' language. The logic and the rules to which it is applied are declared and organized. It is easier to read and map out a decision making process like this, than it is to understand and map out the process buried in other computer programs which are less accessible and where the reasoning and knowledge base are intertwined in a 'quasi-mathematical' notation. 17. Powerful expert systems can run on portable microcomputers (Newton et al., 1988). This means a planning organization can enjoy the benefit of powerful programs on relatively inexpensive hardware. 18. The expertise of an expert system is easily transferred (Waterman, 1985). The transfer of expertise is as easy as copying the program. It could take as little as a few seconds. This is significantly easier than the process involved when one expert tries to pass on her or his knowledge to an apprentice. That process which might involve lectures, books and work experience, can take years. 19. Planners can be involved in the development of the knowledge base, since English-like statements are used to represent the knowledge (Newton et al., 1988). Planners know planning better than knowledge engineers. The opportunity to become WHY SHOULD EXPERT SYSTEMS APPEAL TO PLANNERS? / 37 heavily involved in the process of developing planning tools should increase the likelihood of these tools being useful and truly suited to planners needs. 20. Knowledge about a topic can be accumulated from several sources and experts. By accumulating the knowledge of several sources into its knowledge base, the expert system can work with the benefit of a vast amount of expertise. It can work effectively as a professional colleague for experts themselves, particularly in different specialized niches of one specialty domain. 21. Expert systems promise increased efficiency in the generation of future systems because of their potential for re-using knowledge from existing systems (Miller et al., 1988). Because expert systems can be modified so easily, knowledge bases from other projects can be built upon or altered for different applications. It saves the system developer significant time if it is not necessary to start from scratch. 3.7. CHAPTER SUMMARY Expert systems in theory may benefit urban planning endeavors by overcoming some of the limitations which characterize current computer-based approaches to problem-solving. Much of planning work, by nature, is semi-structured or unstructured. It requires judgment and it is not possible to optimize a single goal. Conventional computer technology to date is geared toward tasks which can focus on a single optimal solution, and which require no judgement or heuristic knowledge. Conventional computer tools, therefore, have been particularly ill-suited to the planning tasks to which they were put in many cases. The predictably disappointing results may account for the distrust many planners hold for computer tools. WHY SHOULD EXPERT SYSTEMS APPEAL TO PLANNERS? / 38 Quantum conceptual leaps into previously unexplored areas is not possible for expert systems and precludes the automation of U t o p i a n type dreaming. Expert systems are no more creative than the processes which are declared for them by human experts. They can work faster and with more information than humans and in this regard can generate conclusions which human experts might not have considered. But the creativity of the system's reasoning process is limited to the creativity of its human expert creators. The rational paradigm has been well served by conventional computer programming. The tendency of expert systems to incorporate softer judgmental information, and accommodate incomplete and uncertain information runs against the rationalist's focus on exclusively scientific, observable and objective knowledge. Expert systems can complement the rationalist approach to problems by addressing some of the perceived weaknesses in the paradigm. Incrementalism is sometimes referred to as the 'science of muddling through.' It describes an approach to planning which is iterative, involves trial and error, and acknowledges that much information is uncertain or incomplete. Expert systems also work this way and may be of more interest to incrementalists than any computer programs in the past Expert systems can combine incremental and rational approaches to planning decision-making. Rational models can be incorporated within an expert system to generate figures or data. The results are then returned to the knowledge-base in the form of symbolic concepts which are manipulated according to the symbolic logic of WHY SHOULD EXPERT SYSTEMS APPEAL TO PLANNERS? / 39 the system. Method planning focuses on procedure. The assumption is that the specific ends are unknown and that only through following established methods which have been found repeatedly successful, will all the necessary information be generated and properly processed. Expert systems may be of assistance if used, for example, as advice givers on the various procedures employed by a planning department. This may be particularly so during complicated and changing situations. Six characteristics of expert systems axe particularly interesting. They complement Langendorfs suggestions about what was required to extend the range of planning problems that could reliably be addressed by computers, and the list of Langendorfs and Lee's criticisms which explain why planners lost faith in early computerized decision-making models. The reasoning of expert systems is accessible and visible to the user; they support an iterative approach to the definition of a problem solving process and resolution of a problem; they can be adapted to changing situations; they can incorporate heuristic reasoning and informal judgmental information; and they do not assume unrealistic quantitative precision. Other features of expert systems which planners might find appealing include: the potential for training novices or helping staff upgrade their skills e.g. in the face of zoning changes; the possibility of accumulating expert knowledge from several different sources; the instant availability of rare and valuable knowledge eg. that used in emergency situations; the increased efficiency and consistency in processing tasks; the relieved burden on experts which would allow them to pursue the more creative WHY SHOULD EXPERT SYSTEMS APPEAL TO PLANNERS? / 40 aspects of their work; the distribution of high-level expertise and skill; and the accountability of expert systems. CHAPTER 4 . WHY EXPERT SYSTEMS MIGHT NOT APPEAL T O S O M E PLANNERS 4 . 1 . INTRODUCTION It would be misleading to suggest that there are no potential drawbacks or concerns related to the introduction of expert systems into planning. There are good reasons to keep people involved in much of the work done by expert systems and it is important to stress what planners should not expect from them. With a good understanding of what the technology can and cannot do, it can be evaluated more honestly and put to work more effectively. This chapter itemizes specific limitations of expert system technology and discusses some general concerns. 4.2. SPECIFIC LIMITATIONS OF EXPERT SYSTEMS Table 4 lists many of the limitations of expert systems, as identified in the literature (Waterman, 1985; Miller et al., 1988). Table 4. LIMITATIONS OF EXPERT SYSTEMS 1. Expert systems are uninspired while people can be more creative. 2. Expert systems must be told what to do while human experts can adapt or learn without being told. 3. Expert systems require symbolic input while human can interpret complex sensory information directly (eg. taste, smell, touch, see). 4. Expert systems have a narrow focus but human experts can have much broader capabilities. 5. Expert systems do not have common sense. 6. Expert systems do not know when they are at the limits of their understanding. 7. Expert systems cannot be left alone to perform a task. 41 WHY EXPERT SYSTEMS MIGHT NOT APPEAL TO SOME PLANNERS / 42 1. Expert systems tend to be uninspired while people can be more creative (Waterman, 1985). They are restricted by the problem solving creativity of the experts and knowledge engineers who built the system. People must be retained to work in conjunction with expert systems so that the strengths of each may be brought to a particular workplace. 2. Expert systems must be told what to do while human experts can adapt without being told. Human experts can learn but expert systems have not come close to matching this ability. They cannot learn to improve their performance on the job as a human expert might. Computer learning is an ongoing area of research (Miller et al., 1988). 3. Expert systems require symbolic input while human expertise can interpret complex sensory experience directly. Visual, auditory, tactile and olfactory input must be translated into symbols if an expert system is to make use of them and significant information can be lost in the process. 4. Expert systems have a narrow focus, whereas human experts can have a much broader focus (Miller et al., 1988; Waterman, 1985). Humans can look at the bigger picture and how other aspects relate to the central issue. Expert systems can often ignore relevant but separate issues because they are focussed exclusively on the problem itself. So much expertise and memory is required to handle simple problems that to handle all the countless potential tangential issues becomes overwhelming. 5. Expert systems do not have common sense knowledge. This goes even beyond tangential issues. Humans have a great deal of common sense knowledge about how the world runs. The following example illustrates this point If a WHY EXPERT SYSTEMS MIGHT NOT APPEAL TO SOME PLANNERS / 43 medical history of a patient listed a weight of 14 lbs and an age of 110 years, most people would assume that the figures had been accidentally reversed. An expert system would not catch this error, however, unless it had been provided with tables of likely age/weight ratios to check against the data (Waterman, 1985). 6. Expert systems do not always know when they have reached the edges of their understanding of a problem (Miller et al., 1985). They do not always know when something is beyond their capacity. In other words they do not know what they do not know. They may try to solve a problem which anyone with common sense would not have wasted time trying to figure out 7. Expert systems cannot be left alone to run autonomously for long periods. Expert systems are interactive and need information from the user frequently throughout a session. A knowledgeable user would know to avoid using the system for a problem which clearly extended beyond its expertise. The above seven points underscore the fact that expert systems cannot work alone. Significant human problem-solving skills have not been replicated by expert systems. If people were not kept in the problem solving process when expert systems were brought in, any advantages of the programs might easily be outweighed by negative consequences of human absence. 4.3. GENERAL CONCERNS REGARDING EXPERT S Y S T E M USE There are some legal, ethical, human and professional concerns starting to appear in the literature. They cannot be adequately explored in this thesis but are introduced briefly to the reader as issues to be followed over the next few years. It is important that these issues be discussed and resolved at some point WHY EXPERT SYSTEMS MIGHT NOT APPEAL TO SOME PLANNERS / 44 4.3.1. Legal As expert system use becomes popular, legal liability problems may arise concerning negligence or incorrect advice. Not all professionals in planning (e.g. ecology versus design specialists, academics versus practitioners) are subject to the same legal and ethical obligations (Wigan, 1987). A qualified professional may use an expert system as a colleague but be able to override what seems to be inappropriate advice in an unusual situation. A non-expert might simply use the advice unquestioningly. Where then might the blame lie if the advice led to regrettable consequences? Expert systems can be designed to offer acknowledgement, references and proper validation for any decision-making advice (Newton et al., 1986). Traditional caveats such as the one listed below may also be written onto the program (Wigan, 1987, p.309). "...Whilst the program has been thoroughly checked and tested on various computers using various compilers, the Standards Association of Australia publishes this program on the basis that neither SAA, its officers nor its committee members are to incur any responsibility or legal liability whatsoever (including liability for negligence) should the contents of the program be incorrect, incomplete or in any other way defective. All liability arising out of or in connection with this program is disclaimed." (Standards Association of Australia, 1986) Unfortunately, exclusion clauses in this type of situation may not yet have been adequately tested by consumer legislation. Given that the main purpose of expert systems is to give advice, and that negligence may be applied irrespective of contracts, it is unproven that such clauses will be effective protection. The 'advantage' of being able to modify expert systems may also raise interesting liability implications for the user. The updated system is no longer the product that WHY EXPERT SYSTEMS MIGHT NOT APPEAL TO SOME PLANNERS / 45 was originally sold by the author, and yet a system which couldn't be modified would quickly become obsolete. 4.3.2. Ethical Issues There is a well documented trend for experts to resist declaring their knowledge for knowledge engineers in the development of expert systems (Wigan, 1987). This raises the extremely difficult issue of intellectual property. To how much of an employee's knowledge does an employer have rights when s(he) intends to develop an expert system and sell it? Can the public or an employer expect an expert to give up their hard earned knowledge which has given them a competitive edge? What is suitable compensation? Another issue which arises in the field of planning, is that the author or firm who builds an expert system, is likely to have their own values and view of how the job is to be done. The purchaser who chooses to buy a particular system also chooses that system authors' view of the problem domain. The system must be used with some appreciation, therefore, of the author's professional reputation and perspective. 4.3.3. Human Resistance to Change Changes in technology are usually, if not always, accompanied by organizational change (Newton, 1986). This may manifest itself in changes in the organization of planning offices, salary scales, communication channels between planners and the public, developers, or politicians. It may significantly alter forms of development control, and even the pace of work. Many such changes are difficult to predict Others, such as initial job displacement, are easier to foresee. WHY EXPERT SYSTEMS MIGHT NOT APPEAL TO SOME PLANNERS / 46 Resistance to change is bound to occur - it may arguably be a part of human nature. Regardless of whether it is justifiable or not, the emotional friction of accepting and adapting to organizational changes can be difficult and unpleasant Some changes may be welcomed, particularly if the tools which precipitate them are easy for most people to use and accept or promise to save lives, for example. Expert systems may be the catalyst for both types of change. 4.3.4. Appropriate Context In the rush to make use of promising new technology, some planners fear that the profession may forget some important wisdom (Sawicki, 1985). People have generally been superior to computers in dealing with a number of major planning aspects, particularly those involving social or political sensitivity. While expert systems promise an impressive expansion of computer capability, they do not possess true 'intelligence' and cannot come close to competing with human experts in certain situations. The planner's ability to recognize the appropriate context for using expert system technology, therefore, remains crucial. Possible problems in this regard would tend to reflect failure on the part of the user rather than tool, but it is a problem to be reckoned with nonetheless. 4.4. CHAPTER S U M M A R Y Expert systems are not a panacea. While they have many advantages to offer planners, there are also disadvantages to be considered. The decision to automate (partially, completely, or not at all) a given task with expert system technology involves considering all the trade-offs. WHY EXPERT SYSTEMS MIGHT NOT APPEAL TO SOME PLANNERS / 47 The more specific limitations of expert system technology include the following. Expert systems are not as creative as people, they cannot learn and must be told what to do, they require symbolic translations of experience which humans can perceive directly (visual, auditory, tactile and olfactory), they have a much narrower focus than humans, they do not possess common-sense like humans do, they do not know when they are at the limit of their understanding, and they cannot be left unattended to work for long periods of time. Other important issues to be considered cover a range of legal, ethical, human and professional interest Expert systems are designed to give advice. If incorrect or inappropriate advice causes negative consequences, liability issues may arise. The scene can be complicated by whether or not the system had been modified, whether or not the user was an expert or a non-expert user. The question of intellectual property is a difficult ethical question to be resolved. When and to what depth do employers have claim to the knowledge of their employees? The changes which this new technology may stimulate in the workplace, the organization, and the pace of work could be resisted by some. There are likely to be both good and bad changes, and these will vary according to a person's perspective. Human adaptation to such changes can be difficult Professional planners have a responsibility to understand what is an appropriate application of the technology and what is not Recognition of the appropriate context for expert system use is crucial if the tools are to be used effectively. Inappropriate applications would not necessarily reflect failure on the part of the tool but would WHY EXPERT SYSTEMS MIGHT NOT APPEAL TO SOME PLANNERS / 48 more likely reflect failure on the part of the user. Expert systems may be used to the most advantage as decision-making aids rather than the final word on a given subject They do not pretend to solve urban planning problems by themselves and the limitations of expert systems must be kept in mind in order that they be used effectively. CHAPTER 5. IDENTIFYING TASKS IN PLANNING WHICH W O U L D SUIT EXPERT S Y S T E M T E C H N O L O G Y 5.1. INTRODUCTION It is important to know that the expert system approach, as a work tool, is suited to the problem at hand. This improves the likelihood of obtaining useful results and avoids an unnecessary waste of time and resources in the decision-making process (Ortolano et al., 1987). The first section of this chapter suggests guidelines to consider when evaluating a task's suitability for expert system application. The final section describes different types of expert systems and then suggests possible planning applications across the spectrum. The suitability guidelines reflect the cumulative wisdom of Han and Kim (1989), Newton and Taylor (1988), Ortolano and Perman (1987), and Waterman (1985). The guidelines are a logical response to the benefits expert systems can bring to certain planning tasks, the disadvantages of which planners should be wary, and the practical limitations of distinguishing between those tasks that can be effectively translated into expert system code and those that cannot 5.2. GUIDELINES FOR DETERMINING TASK SUITABILITY T O EXPERT SYSTEMS Certain conditions are often recommended for determining the suitability of a given task to expert system technology. The first of these conditions simply calls for an awareness of existing technology at planners' disposal. There may be effective conventional software already available. The second of these conditions involves an 49 IDENTIFYING TASKS IN PLANNING WHICH W O U L D SUIT EXPERT SYSTEM T E C H N O L O G Y / 50 appreciation of the advantages and disadvantages implied in expert system technology. The remaining six conditions affect the ease with which the decision-making process of that problem domain can be effectively encoded into an expert system. (i) The task involves heuristic (non-numeric, non-algorithmic) knowledge and cannot therefore be satisfactorily handled by conventional software. If the job can be done by existing programming, it is doubtfully worth investing more time and money in an expert system. If the task involves heuristic knowledge then it cannot be handled by conventional software. (ii) The benefits are worth the trade-offs and investment The benefits expected from the expert system must be considered worth the time and money investment that is required to develop the system. They must also be considered worthwhile in the face of the disadvantages or limitations of expert system technology. Does increased productivity or consistency at a particular job in the planning office outweigh the loss of discretion or creativity, for example? These factors may not be those in question for a particular example, but there may well be trade-offs to be considered. It is here that the potential advantages and disadvantages discussed in Chapters 3 and 4 become important (iii) The knowledge domain for the task at hand is specialized and the problem can be clearly defined within boundaries. Expert systems have a limited capacity for broad fields of knowledge or general common sense. Planning can not be developed into an expert system. Only limited tasks within planning which can be well defined within boundaries, can be effectively codified. (iv) The task or decision-making process is not poorly understood (Han et al., 1989). IDENTIFYING TASKS IN PLANNING WHICH W O U L D SUIT EXPERT SYSTEM T E C H N O L O G Y / 51 If the process for handling the task is poorly understood, then it is difficult to declare and encode that process for the expert system's use. The process does not have to be perfectly understood because there is room for trial and error in the development process. Rules can be modified until the outcome is the desired one, but beginning with a poorly understood process may be stretching this flexibility to an extreme. The better understood the process, the easier is the task to encode. (v) True 'experts' actually exist, meaning it is acknowledged that those called experts can do the job better than novices. By definition, expert systems are best suited for specific areas of expertise. These are easier for expert systems to handle because they require less knowledge than more generalized tasks. Common sense is virtually impossible to encode because it involves a vast base of knowledge. If a novice can perform a task as well or as poorly as an expert, it is probably on the basis of common sense. There may not be any refined knowledge or expertise for that task. (vi) The task at hand is generally complex enough to take more than fifteen minutes and less then several hours of an expert's time to perform (Silverman in Han et al., 1989). This is a very rough guideline. It is generally considered to be a reasonable range of complexity for encoding purposes. A task which takes an expert less than fifteen minutes to do may be too simple to be worth developing a system, and one which takes days or weeks to solve may be too complex to encode. The frequency with which a task is performed should be considered at the same time, however. If something takes days to perform but is routinely done, it may still be worth the time to try encoding it no matter how complex. The job can be broken down into simpler sub-tasks for system development purposes. IDENTIFYING TASKS IN PLANNING WHICH W O U L D SUIT EXPERT SYSTEM T E C H N O L O G Y / 52 Similarly, if a task such as responding to a routine enquiry takes only ten minutes but experts are distracted from more difficult work by having to perform it constantly throughout the day, it may be worth developing an expert system. A less experienced employee could respond to the enquiries with the aid of the expert system, only bothering the expert for unusual cases, (vii) An articulate expert is ready and willing to cooperate with the system developers until the system is complete. Cooperation includes declaring one's decision-making process, and critiquing the system prototypes throughout the development process (possibly 2 years). The time an expert may wish to take out from other work, and personal objections regarding intellectual property are both factors which may affect an expert's willingness to cooperate fully. (viii) Progress can be made incrementally. This has proven to be the most effective way of developing expert systems. A general framework of the rules governing a decision making process is encoded. The system is then run repeatedly with examples of typical problems. Raws and omissions in the reasoning process are gradually identified and corrected through successive runs until the program operates as smoothly as would an expert at that task. If a task can be improved in small steps this way (by adding and changing rules), then the development of an expert system is greatly facilitated. 5.2.1. Discussion Initially, some of the eight criteria above may appear problematic in a planning context The existence of true experts in planning may be disputed at the general level because, for example, there is no collectively agreed upon way to correctly plan a city. Typically expert systems have been conceived for highly specific problem areas rather than multifaceted professions such as planning which embraces social, economic IDENTIFYING TASKS IN PLANNING WHICH W O U L D SUIT EXPERT SYSTEM T E C H N O L O G Y / 53 and even political factors (Han et al., 1989). Planners of different philosophical stripes may have difficulty agreeing upon solutions or even approaches to solving a problem in many cases. Finding well-bounded planning problems which take between a few minutes and a few hours to solve, and which are well understood, can also be a challenge. Political processes and the opposing goals of many groups with different values, often work to complicate issues drawing out resolution for months or years. It has been suggested that because "... few planning problems satisfy these conditions, the role of expert systems in planning will be rather limited, and the development of fully operational systems will be slow and arduous" (Klosterman and Landis, 1988, p.362). The conditions to which Klosterman and Landis were referring were: appropriate tasks are small and well defined; experts perform better than non-experts; conventional programs are inadequate for the task; and an articulate and cooperative expert is available for the long term for system development Indeed, these same authors seem doubtful that expert systems will ever experience widespread use in the field of planning, in spite of their oft-touted potential. On the other hand, there are researchers of the opinion that there are numerous tasks within the planning domain which are suitable to expert system use (Ortolano et al., 1987). There is certainly no question but that attempts will be made to develop urban and regional planning applications. Many are already in progress. Reasons for which many planners have expressed interest in expert systems are IDENTIFYING TASKS IN PLANNING WHICH W O U L D SUIT EXPERT SYSTEM T E C H N O L O G Y / 54 outlined in Chapter 3. They include the fact that expert systems can handle uncertainty; can be adapted to changing circumstances; and can use heuristic decision-making wisdom that could not be accommodated in conventional computer programs. While planning on the whole cannot be neatly described in terms of the eight suitability criteria outlined in this section, it bears mentioning that nor can planning on the whole be described in the numerical, rational or algorithmic terms ideally necessary for the application of most conventional computer programs. Nor can contemporary planning be described in terms which would suit it ideally to non-computerized human problem solving techniques. The amount of information being processed, the scale of work in modern day cities and the expectations and quality of service demanded of planners in this day and age require sophisticated information and time-saving technology. Expert systems may not represent the perfect tool for planners but they may yet prove to be better in many ways and in certain circumstances than the tools (computer-based or otherwise) currently being used. Planning tasks may not match up perfectly with the conditions academically preferred for expert system use, but they may match more of these conditions in certain cases, than they match on other lists for other 'tools.' It of interest to note that there are many planning situations in which conventional computer programs are working satisfactorily, whether they be in data-base management, spreadsheet applications, word processing, or technical modelling. They may IDENTIFYING TASKS IN PLANNING WHICH W O U L D SUIT EXPERT SYSTEM T E C H N O L O G Y / 55 be doing the job perfectly, or they may be performing with recognized limitations but better than could be done manually. The successes of conventional programming illustrates the possibility that while planning in general does not meet the criteria necessary to benefit from a particular tool, smaller tasks and functions within the larger picture can meet such criteria. The larger problem of planning may be broken down into more manageable parts. The manner and strategy by which this decomposition occurs may or may not be appropriate to expert system technology. Smaller parcels may have a greater likelihood, however, of benefiting from the expert system tool. If expert system developers shy away from planning problems because they are more complicated and less structured than other domains (such as medicine or chemistry), planners may take advantage of the technology anyway. It could represent an imperfect but preferable decision-making alternative in many contemporary planning situations. 5.3. EXPERT S Y S T E M P R O B L E M TYPES Expert systems have met with success in certain main areas of application and can be thought of as representing different generic problem solving types. There is no formal consensus on a typology. Some researchers suggest that there are only two or three problem types, while others identify several more (Miller et al., 1988). They can help the user to understand the types of problems, which are suitable to expert system use, regardless of the specific knowledge domain. Reflecting the more moderate centre of the spectrum, one typology is outlined below. IDENTIFYING TASKS IN PLANNING WHICH W O U L D SUIT EXPERT SYSTEM T E C H N O L O G Y / 56 Planning applications may fall within any one or a combination of these categories but (ii), (iii), and (v) seem to be the most likely areas of interest (i) Diagnosis and repair. (ii) Design and planning. (iii) Interpretation. (iv) Monitoring and control. (v) Instruction. 5.3.1. DIAGNOSIS AND REPAm Problems in this category are characterized by a fixed set of alternatives, question -answer dialogue, and differential weighing (Miller et al., 1988). Classification problems such as helping heritage planners to classify a building's architectural style and historic period, might be an interesting example of this type. Alternatively, many diagnosis expert systems may be developed to infer malfunctions from observable data and prescribe remedies for the situation. Local governments concerned with public works might find this to be of interest Relatively few planning applications are envisioned for this category (Ortolano and Perman, 1990). 5.3.2. DESIGN AND PLANNING In the area of design and planning, key characteristics of the system would be the open-ended space of alternatives, and the coordination of multiple experts on a subject (Miller et al., 1988). The expert systems suggest forms of arrangements of objects and actions under given constraints. Potential applications would include site analysis, land-use location, and environmental planning activities. Mathematical models are commonly used in urban and regional planning and expert systems could help planners determine which mathematical models should be used in a given situation (Ortolano et IDENTIFYING TASKS IN PLANNING WHICH W O U L D SUIT EXPERT SYSTEM T E C H N O L O G Y / 57 al., 1990). 5.3.3. INTERPRETATION Interpretation applications refer to those situations where descriptions must be inferred from data. Examples might include whether a proposed land-use meets zoning and other local land-use regulations. Expert systems providing assistance in the analysis and use of land-use laws would be another such application. 'Intelligent databases' could also be useful although there are none operational in planning yet The user of a database must usually stipulate parameters for a search and is responsible for her or his own interpretation of the data. An expert system data-base would help a user use a data-base more effectively. Having established what the user wants, the expert system could define the parameters and interpret the data for the user. 5.3.4. MONITORING AND C O N T R O L Expert systems in this category are those which compare observations to expected outcomes and govern the behaviour of the monitored systems themselves. The only example found of this type of application which was close to planning, was a system which controlled the environmental conditions in buildings. This could lead to interesting possibilities for managing and studying traffic flows in transportation planning. Environmental planners may also find it useful to develop systems which monitor and interpret soil air and water quality (Miller et al., 1988). IDENTIFYING TASKS IN PLANNING WHICH W O U L D SUIT EXPERT SYSTEM T E C H N O L O G Y / 58 5.3.5. INSTRUCTION The final category covers those expert systems which help novices with new concepts or train people to perform new tasks. New staff, for example could be taught about complex zoning requirements or development control laws (Ortolano et al., 1987) The interactive and explanatory features of expert systems are advantages in a training situation. 5.4. PLANNING APPLICATIONS Numerous planning expert systems of different kinds are currently being researched and developed. Table 5 makes brief reference to some of these and should give the reader an idea of the breadth of potential planning application. Subsequent chapters (6,7 and 8) elaborate on three of these examples (those marked with an asterisk). IDENTIFYING TASKS IN PLANNING WHICH WOULD SUIT EXPERT SYSTEM TECHNOLOGY / 59 Table 5. SAMPLE PLANNING APPLICATIONS OF EXPERT SYSTEMS AUTHOR OR SYSTEM NAME DESCRIPTION MEDIATOR Lee I Wiggins (Kim, Wiggins t Wright 1990) An expert system t o f a c i l i t a t e environmental dispute r e s o l u t i o n . * Plan Checker (E.B. Economics, 1990) A s s i s t s P lan Checkers at C i t y H a l l (Vancouver) check plans f o r s p a t i a l s eparation requirements. Necessary p r i o r t o approval of development permits. PROSPECTOR (Han t Kim 19B9) Aids g e o l o g i s t s i n search f o r ore d e p o s i t s . R i t c h i e Mahonie 1986 (Kim, Wiggins t Wright 1990) An expert system t o a s s i s t with highway, sever and bridge maintenance. RTKAS (Kim, Wiggins t Wright 1990) An expert system f o r r e a l - t i m e monitoring and a n a l y s i s of t r a f f i c during evacuation. SACON Bennet et a l . 1978 (Ortolano t Perman 1987) A s s i s t s i n s e l e c t i n g models for s t r u c t u r a l a n a l y s i s . • SCREEKER (ESSA L t d . 1990) A s s i s t s environmental o f f i c e r s of Transport Canada A i r p o r t A u t h o r i t y i n screening c a p i t a l p r o j e c t s f o r environmental impact assessment. SISES F i n d i k a k i ' (Ortolano t Perman 1987) S i t e s e l e c t i o n f o r a p a r t i c u l a r land-use. URBYS Tanic 1986 (Klosterman 1 Landis 1988) Developed f o r Fort-de-France, M a r t i n i q u e . Prototype combining expert system w i t h database management system. Tay l o r 1986 (Newton 1966) Impact of s t r e e t c l o s u r e on neighborhood road network. IDENTIFYING TASKS IN PLANNING WHICH WOULD SUIT EXPERT SYSTEM TECHNOLOGY / 60 Table 5 (Continued) SAMPLE PLANNING APPLICATIONS OF EXPERT SYSTEMS » AUTHOR OR SYSTEM NAME DESCRIPTION AIRPLANE Carnegie-Mellon Oniv. ( M i l l e r and Walker 1988) A i r c r a f t scheduling, t r a f f i c -c o n t r o l , route planning. AT T B e l l Lab (Ortolano 1 Perman 1987) System i n t e r p r e t s numerical outputs f o r users of standard s t a t i s t i c a l a n a l y s i s computer programs. CHINA ( H a r r i s , Conn, Bowlby 198?) Computerised Highway Noise A n a l y s t . A s s i s t s i n the ac o u s t i c design of highway noise b a r r i e r s . Development Control Expert System (Leary 1988) Helps l o c a l planners t o implement development c o n t r o l s . Design g u i d e l i n e s based upon Essex County Design Guide. ESSAS (Han 1 Kim 1987) Determines s u i t a b i l i t y of s i t e f o r c o n s t r u c t i o n of m i l i t a r y f a c i l i t i e s . ESMAN (Han I Rim 1986) An expert system f o r manufacturing s i t e s e l e c t i o n . Expert Systems t a t . (Ortolano a Perman 1987) Provides advice on b u i l d i n g r e g u l a t i o n s . Cere and Coyne 1986 (Rim Wiggins I Wright 1990) Expert systems combined w i t h Computer Aided Design (CAD). Crant Advisor ( r i l e 1988) Advises users whether or not they q u a l i f y for municipal grants and programs. * HERMES (A l b e r t a Research Council A l b e r t a P u b l i c Safety S e r v i c e s ft Emergency Preparedness Canada) H e u r i s t i c Emergency Response Management Expert System. Provides advice for a c c i d e n t s i n v o l v i n g hazardous s p i l l s . L a n d f i l l S i t e S e l e c t i o n Rouhani a Kangari, 1967 (Kim, Wiggins t WRight 1990) An expert system for s e l e c t i n g s i t e s f o r l a n d f i l l s . Law, Simaie Chapman 198* (Ortolano fc Perman 1987) Expert system to c h a r a c t e r i s e i n a c t i v e hazardous waste d i s p o s a l areas. S i t * a n a l y s i s . IDENTIFYING TASKS IN PLANNING WHICH W O U L D SUIT EXPERT SYSTEM T E C H N O L O G Y / 61 5.5. CHAPTER SUMMARY The intent of this chapter was to develop guidelines from which to judge a planning task's suitability for expert system development The eight guidelines were:(i) the task involves heuristic reasoning and judgment and therefore cannot be helped by conventional computer technology; (ii) the benefits are worth the trade-offs and investment; (iii) the knowledge for the task at hand is specialized and well bounded; (iv) the decision-making process is well understood; (v) true experts exist and can do the job better than novices; (vi) the task takes human experts generally between fifteen minutes and several hours to perform; (vii) an articulate expert is willing to cooperate until the expert system is complete; and (viii) progress can be made incrementally. These guidelines were based upon the cumulative suggestions of Han and Kim (1989), Newton and Taylor (1988), Ortolano and Perman (1987), and Waterman (1985). They reflect the expert system benefits of which planners should try to take advantage (discussed in Chapter 3), as well as the drawbacks (Chapter 4) and encoding limitations. Tasks which cannot be properly encoded into an expert system cannot be performed well by that system. Planners' initial reaction to the eight suitability guidelines may be that, because planning cannot be described in all the terms listed, expert systems technology is ill-suited to planning. It may be too hasty to jump to such conclusions. It should be remembered that while in the larger picture, planning may not meet all of the stated criteria, many of the tasks within the larger picture do. IDENTIFYING TASKS IN PLANNING WHICH W O U L D SUIT EXPERT SYSTEM T E C H N O L O G Y / 62 The necessity to evaluate the particular job within planning must be emphasized (Ortolano et al., 1987). There are areas where it may be a waste of resources to develop expert systems, but there are many areas where expert systems may be far more worthwhile than conventional computer or manual methods. Specific suitability guidelines reflecting liability and professional ethical issues were not developed because these issues have been inadequately explored in the literature. Expert system performance in planning over the next few years and the problems or successes that ensue will help us pursue these ideas. For now, planners may wish to incorporate their own views in these areas when they consider the second guideline, whether or not the benefits are worth the trade-offs and investment for that particular task. A typology of five generic problem-solving types may help the user understand the nature of problems for which expert systems can be useful. The categories are: (i) diagnosis and repair; (ii) design and planning; (iii) interpretation; (iv) monitoring and control; and (v) instruction. Planning examples may fall within any one of the categories or be a combination of two or more. Categories (ii), (iii) and (v) may be the areas of most frequent interest to planners. A table outlining examples of planning applications was introduced to give the reader an idea of the breadth of planning tasks which have been considered suitable for expert system development Most of these are still being built although some may be complete by now. Examples included expert systems to help local planners in Britain implement development controls (Leary, 1988); to find alternative sites for refuse transfer and disposal (Kim et al., 1990); to estimate the impact of street closure on a IDENTIFYING TASKS IN PLANNING WHICH W O U L D SUIT EXPERT SYSTEM T E C H N O L O G Y / 63 neighborhood road network (Newton, 1986); to interpret numerical output for users of standard statistical analysis computer programs (Ortolano et al., 1987); and MEDIATOR, to facilitate environmental dispute resolution (Kim et al., 1990). Chapters 6, 7 and 8 will discuss three Canadian examples of expert system planning applications in more detail. CHAPTER 6. HERMES ILLUSTRATIVE APPLICATION 6.1. HERMES - ORIENTATION 6.1.1. Developers The Heuristic Emergency Response Management Expert System (HERMES) is an example of expert system technology applied to emergency planning. The system was developed by the Alberta Research Council (ARC), Alberta Public Safety Services (APSS), and Emergency Preparedness Canada (EPC) (Chang et al., undated). 6.1.2. Function HERMES has been developed as a prototype to ascertain the feasibility of designing an "intelligent assistant and tutor" for emergency response management personnel (Clark et al., 1988). The system offers advice and assistance in the management of a crisis involving hazardous materials. On the basis of information supplied by the user, it can estimate hazard levels such as fire and explosion risk, recommend procedures for materials containment or human evacuation, request important information which has not yet been supplied, and perform various other advisory functions. Internal calculation models and heuristic rules can be applied to users' input in order to determine such things as dispersal plumes, the distance of projectiles upon explosion, and evacuation areas (Gaussian plume equation (Chang et al., undated). The system digests all the available information, and advises remedial action using the same logical processes as would human experts (Clark et al., 1988). Final decisions on remedial action are taken by the onscene commander who combines advice received from the 64 HERMES ILLUSTRATIVE APPLICATION / 65 centre with her or his own knowledge of the situation. 6.1.3. Users HERMES is intended for use by emergency management personnel, many of them experts themselves in one disaster area or another, and others possibly less experienced. There is also the HERMES Tutorial system which is designed for use by trainees in this field. Typically in Alberta, police and firefighters faced with a dangerous goods accident will contact the province's Compliance Information Centre (CIC) 24-hour hotline for advice. The human experts at the centre provide emergency management direction, advising how to control the problem, how to mitigate long-term impacts, and how to clean up the situation as safely, effectively and cheaply as possible. The system can be thought of as an expert assistant A human expert who is unfamiliar with a given type of emergency could consult with HERMES to quickly determine an appropriate course of action. The system would be helpful to new personnel, offering quick and sound advice until more experienced experts could be contacted and their help solicited. HERMES would also help highly experienced experts, offering confirmation and acting as a professional colleague with whom to consult for important decisions in urgent times. HERMES ILLUSTRATIVE APPLICATION / 66 6.1.4. Scope The version of HERMES discussed in this case-study was designed as a proof-of-concept prototype and was, therefore, kept fairly simple. The knowledge base has been assembled to handle only a few hazardous materials generally encountered in surface transportation. The prototype has since been judged successful and the next phase of development, namely the construction of a more useful and operable system, has just begun. The next version of HERMES should be capable of working with a full range of known hazardous materials, and handling more complicated situations (Chang et al., undated). 6.1.5. Technical The HERMES package was developed on the Symbolics 3620 LISP machine. It can be run on the same, or an IBM 386 machine. The software used to develop and run HERMES was the Automated Reasoning Tool (ART) by Inference Corporation. ART offers both forward and backward chaining. In forward chaining, the system proceeds forward to a conclusion or suggested procedure, such as generating advice in HERMES. In backward chaining, the system works backward from a hypothetical conclusion, and tries to fill in the missing information to understand why something occurred. This can help HERMES devise questions for the user (Clark et al., 1988). The system asks only for the specific information it needs, and does not waste the time of on-scene personnel by asking for related information which is ultimately not used. HERMES ILLUSTRATIVE APPLICATION / 67 6.2. P R O B L E M SUIT ABILITY 6.2.1. Characteristics The work done by personnel in the Compliance Information Centre is primarily to determine remedial action according to the specific conditions of a given emergency. It involves both immediate and post-crisis action. The former calls for identification of the materials in question, risk to human life, risk of fire and containment of the spill. Post-crisis action involves clean-up procedures and an evaluation of possible longterm effects on the environment or people (Chang et al., undated). The primary focus of emergency planning is to save human life and prevent suffering. The protection of the environment and property, containment and clean-up of the spill become the major focus once danger to human life is past Upon receiving a call from police or firefighters, the expert on duty at the CIC obtains a description of the situation. The expert must then draw upon information from a computerized database, reference manuals, reports, and personal experience to fully understand the problem and to offer sound advice on how to cope with it There have been difficulties associated with the work: • Large amounts of information must be searched and many factors balanced. During a crisis, stress and the accelerated pace of work can cause even well-qualified personnel to overlook things. There are countless types of hazardous materials. The volume of materials-specific information that could be useful far exceeds the memory capacity of human experts. HERMES ILLUSTRATIVE APPLICATION / 68 • The often incomplete and uncertain nature of information being received during an accident can make it difficult for the expert who is not at the scene to offer quick and useful advice. • The time, place, environmental conditions, and nature of an accident are unpredictable. It is impossible to ensure that an appropriate expert is on duty 24 hours a day, and 365 days of the year. • If the expert on duty has no direct experience with the accident type, procedures for that particular example may not have been outlined in resource manuals, and delays can result as more expertise is located. In the alternative, unnecessary errors may be made if the expert on duty must give advice before more knowledgeable experts have arrived. • Information access can prove tedious and experience is necessary to conduct an effective search. 6.2.2. Suitability Criteria The type of work done by emergency response management personnel matches the guidelines suggested in Chapter 5.0 for estimating the potential suitability of expert systems to a given task. For the reader's convenience these are outlined again in Table 6. Each is then briefly discussed in the context of this application. HERMES ILLUSTRATIVE APPLICATION / 69 Table 6. EMERGENCY RESPONSE MANAGEMENT MEASURED AGAINST TASK SUITABILITY CRITERIA C r i t e r i a S a t i s f a c t i o n i ) Task invo l v e s h e u r i s t i c knowledge and cannot be handled by conventional systems. i i ) B e n e f i t s are considered worth the investment and co s t s . T i i i ) S p e c i a l i z e d knowledge domain and c l e a r problem d e f i n i t i o n . y i v ) Task process not poorly understood. y v) True experts e x i s t and perform b e t t e r than novices using commons sense. Y v i ) Task takes 15 min. to sev e r a l hours t o perform (without expert system). (Y) v i i ) A r t i c u l a t e expert i s w i l l i n g to cooperate u n t i l system complete. y v i i i ) P r o g r e s s can be made increm e n t a l l y . LEGEND: Y « Yes, c r i t e r i a i s s a t i s f i e d . (Y) • C r i t e r i a i s s a t i s f i e d , more often than not. (i) The task involves heuristic knowledge (non-numeric, non-algorithmic) and cannot therefore be handled by conventional computer programming. Yes. The tasks involve both numeric and heuristic decision-making. While some mathematical models and equations are used where considered appropriate in sub-tasks, the system relies on rules of thumb, shortcuts and experience supplied by experts. (ii) The benefits are worth the investment Yes. The benefits anticipated from the use of HERMES seem to outweigh the drawbacks as far as the groups who are funding the project are concerned. Table 7A and 7B indicate the many benefits which could be anticipated for this type of expert system application. Table 8, indicates those expert system limitations which would be of most, concern for emergency response planning. HERMES ILLUSTRATIVE APPLICATION / 70 (iii) The knowledge domain for the task at hand is specialized and narrowly focussed, and the problem can be clearly defined. Yes. The domain knowledge for emergency response planning is vast but specialized in terms of the characteristics of various chemicals. Its focus is also fairly specific namely, the preservation of human life, the containment and clean-up of spilled hazardous materials, and mitigation of adverse long term effects where possible. There are many indirect and invisible effects to some hazardous materials accidents, but in the main, the problem can be clearly specified. (iv) The task or decision-making process is not poorly understood. True. While the decision-making process is not perfectly understood nor rigidly defined, there are recognized procedures and priorities for handling accidents involving hazardous spills. There is also room for flexibility and creativity in the process because no two accidents are the same. When dealing with dangerous materials, decision risks are necessarily kept as conservative as possible. Appropriate basic response processes for certain accident types exist but advice at a detailed level would vary greatly depending upon the incident The short and stressful time span for decision-making in emergencies can cause a previously agreed upon process to be forgotten and unnecessarily compromised, however. (v) True 'experts' exist and can do the job better than novices. Yes. True 'experts' undoubtedly exist They cannot claim to be able to solve all the problems in their domain any more than doctors understand all disease, but they are eminently more capable at their work than would be the average person using common sense. There is also professional agreement generally, upon how to approach different types of hazardous materials accidents. This is formally expressed in reference HERMES ILLUSTRATIVE APPLICATION / 71 and emergency manuals. vi) The task at hand is generally complex enough to take more than fifteen minutes and less then several hours of an expert's time to perform. Generally. As information is received, decisions are often requested rather quickly for stop-gap measures, and ongoing emergency management Each of these decision tasks could conceivably fall within the time frame suggested by the guideline. Some of the necessary information required for decisions may take longer than several hours to be obtained, though. The life-threatening nature of these accidents and the fact that seconds count, would suggest that expert systems would be a worthwhile investment, if they could perform faster than humans, regardless of whether they fall strictly within the suggested range or not (vii) An articulate expert is ready and willing to cooperate with the system developers until the system is complete. In this case, a cooperative and enthusiastic expert was available for the development of the system. The expert was not always good at declaring his decision-making knowledge. By discussing different situations and prototyping the system, his process was gradually documented. (Such a process can be an excellent opportunity to critique and improve even an expert's process). (viii) Progress can be made incrementally. Progress can be made incrementally in emergency response management The HERMES prototype (feasibility study) can only handle a select number of emergency types (limited by its knowledge base of only a few common hazardous materials). Expert system technology will allow for easy modification and adaptation to new materials, more complicated situations and more knowledge. HERMES ILLUSTRATIVE APPLICATION / 72 If experts at the CIC were to use HERMES in their work; accidents involving chemicals for which information was not built into the knowledge base would have to be handled from written materials, the data-base, and experience as is currently the case. As the knowledge base was expanded to include a wider range of information, it would gradually be used more and more. HERMES ILLUSTRATIVE APPLICATION / 73 Table 7A POTENTIAL BENEFITS OF USING EXPERT SYSTEMS FOR HERMES PROBLEM (Main Benefits of Interest to Planners) MAIN EXPERT SYSTEM FEATURES OF INTEREST TO PLANNERS 1. Expert systems can expand the range of planning tasks to be automated. 3. 4. 5 . 6. Reasoning a c c e s s i b l e . i s Supports i t e r a t i v e and i n t e r a c t i v e problem s o l v i n g . F l e x i b i l i t y to adapt to changing c o n d i t i o n s . I n c o r p o r a t i o n of h e u r i s t i c reasoning. Does not q u a n t i t a t i v e p r e c i s i o n . assuae IS FEATURE BENEFICIAL FOR THIS PARTICULAR PROBLEM APPLICATION? Tes. Models f o r t h i s type of work i n the past have been d i s a p p o i n t i n g . This task could not be e f f e c t i v e l y automated before. Yes. T h i s feature i s c r u c i a l . Users must have confidence i n system's advice i f l i v e s are a t stake. E x p l a n a t i o n c a p a b i l i t i e s a l l o w CIC s t a f f t o evaluate and p o s s i b l y o v e r r i d e advice, or t o l e a r n something new. Yes. The system's knowledge base can be modif i e d as understanding improves. Yes. For same reasons as above. Yes. Advice and reasoning i n v o l v i n g s h o r t c u t s , judgments and stop-gap measures are v i t a l . Saved time may save l i v e s . The system i s a l s o capable of numerical a l g o r i t h m s where a p p r o p r i a t e . Yes. The incomplete i n f o r m a t i o n l i k e t h i s , w i t h what advice w i t h f a c t o r s . take t h i s i n t o account f i n a l d e c i s i o n s . a b i l i t y to work w i t h and u n c e r t a i n i s necessary to a job I t must be able to work i s a v a i l a b l e and g i v e approximate confidence On-scene personnel can when making HERMES ILLUSTRATIVE APPLICATION / 74 Table 7B. POTENTIAL BENEFITS OF USING EXPERT SYSTEM FOR HERMES PROBLEM (Additional Benefits of Interest To Planners) ADDITIONAL E.S. FEATURES OF INTEREST TO PLANNERS IS FEATURE BENEFICIAL FOR THIS PARTICULAR PROBLEM APPLICATION? 7. Preserve s p e c i a l i s t knowledge. Tes. Rare but valuable knowledge may be forgotten or l o s t by humans. 8. T r a i n i n g p o t e n t i a l . Tes. Personnel can p r a c t i c e innumerable s i t u a t i o n s before being q u a l i f i e d f o r r e a l duty. Q u a l i f i e d s t a f f can p r a c t i c e to keep t h e i r s k i l l s sharp during 'quiet' times. 9. Freeing experts f o r •ore c r e a t i v e work. Tes. C r e a t i v e t h i n k i n g may be v i t a l i n d i s a s t e r s i t u a t i o n s and should be used to complement advice of system. 10. System net bewildered by vast amount of information. Tes. T h i s can save time and lead to more informed d e c i s i o n s . 11. I m p a r t i a l l y explore a l l hypotheses. Tes. 12. I n t e r a c t i o n at each user's l e v e l . Tes. Experts have varying degrees of f a m i l i a r i t y of d i f f e r e n t s i t u a t i o n s . 13. Consistency. Tes. In times of s t r e s s , people can be i n c o n s i s t e n t . 14. Increased p r o d u c t i v i t y . Tes. An expert system can work f a s t e r and handle more things at once. 15. E f f i c i e n c y . Tes. R e l a t i v e l y small amount of input can generate well-foeussed information needs. Asking user f o r missing in f o r m a t i o n , system invokes only the r u l e s s p e c i f i c a l l y needed. 16. Reasoning easy t e document. Tes. Can be u s e f u l f o r reviewing and e v a l u a t i n g responses a f t e r s i t u a t i o n over. Chain of reasoning can be c r i t i q u e d . 17. Runs • on microcomputer. Tes. Large c i t i e s may choose t o run systems on f a s t e r and more s o p h i s t i c a t e d computers. Small towns may be l i m i t e d t o l e s s powerful microcomputers. 18. Expertise e a s i l y t r a n s f e r r e d . P o s s i b l y . I f system i s t o be d i s t r i b u t e d t o other o r g a n i z a t i o n s . 19. Experts can be involved in system development. Tes. 20. Accumulated knowledge from several sources. Tes. The cumulative wisdom of a l l the best minds can be used when catastrophe s t r i k e s . 21. Modified knowledge base u s e f u l f o r future systems. Tes. HERMES ILLUSTRATIVE APPLICATION / 75 Table 8. DRAWBACKS OF USING EXPERT SYSTEM FOR HERMES PROBLEM EXPERT SYSTEM LIMITATIONS OP CONCERN TO PLANNERS IS LIMITATION OF NOTABLE CONCERN FOR THIS PROBLEM APPLICATION? 1. Uninspired. Not c r e a t i v e l i k e people. Yes. For t h i s reason, HERMES must be used as an a s s i s t a n t o n l y . People must be kept as part of process because of i r r e p l a c e a b l e c r e a t i v e problem s o l v i n g t a l e n t s . 2. Cannot adapt, l e a r n without being t o l d . » Yes. For t h i s reason, s t a f f must c o n t i n u a l l y review performance of HERMES and update knowledge base. 3. Cannot i n t e r p r e t sensory data as w e l l as people. No. This i s always a problem when requesting advice from CIC. S t a f f there cannot d i r e c t l y i n t e r p r e t surroundings l i k e on-scene personnel. Must work i n s p i t e of t h i s l i m i t a t i o n . 4. Expert systems have narrow focus. Yes. Myriad of p o s s i b l e accident types may need information from t a n g e n t i a l f i e l d s which HERMES i s not prepared f o r . This can be even more problematic for human experts though, who can remember and manipulate l e s s information at one time. 5. Expert systems have no common sense beyond s p e c i f i c knowledge area. Yes t h i s c ould be a problem. For t h i s reason, people w i l l remain part of the process. HERMES need not be used, or advice need not be heeded i f common sense of s i t u a t i o n d i c t a t e s such. «. Do not know when edges of knowledge reached. Yes. Personnel must be able to recognize a s i t u a t i o n l i k e t h i s . 7. Cannot run autonomously f o r l o n g periods of time. No. The nature of t h i s work i s such that many people are i n v o l v e d and information and advice i s c o n s t a n t l y being exchanged. The system would not be expected t o work alone. An i n t e r a c t i v e system i s p r e f e r a b l e . 6.3. PERFORMANCE HERMES ILLUSTRATIVE APPLICATION / 76 6.3.1. Main Features 6.3.LI. User Interface HERMES has been designed to be user-friendly. Graphics, and several "windows" on the main screen facilitate interaction with the user. Information entry is flexible and can be done with a mouse or keyboard. Even the order of entering information can vary and be done according to the user's usual thinking style (Clark et al., 1988). This is a major advantage of HERMES and helps the user enter information in the order in which it comes and in an order that makes sense to them. Users can retain their own procedures for information collecting and documentation, to a large degree, and not have them determined by the machine. 6.3.1.2. SCENE Window The user communicates the accident scene to HERMES through placing symbols in the SCENE window. Trucks, people, roads, water and so on can be placed graphically in their relative positions. Weather and microclimate considerations can also be indicated in this way, using graphical symbols for wind direction, temperature etc. This same window is where evacuation limits and hazardous materials concentration areas are displayed. HERMES ILLUSTRATIVE APPLICATION / 77 6.3.1.3. SUGGESTED THINGS TO DO window This window lists additional things to do and information to be collected from the scene or elsewhere. 6.3.1.4. ADVISED EMERGENCY ACTION GIVEN window This window offers the emergency advice, listing procedures and actions to be taken immediately. The advice appears after all input information has been processed. 6.3.1.5. WATCH THESE LEVELS window Various gauges are displayed in this window, reflecting risks along various dimensions. There are gauges for evacuation danger, situation danger, explosion danger, fire danger, site clanger, public danger, worker danger, environmental danger, and so on. The readings of these gauges should decrease if all advice given is implemented. The worker danger gauge would decrease for example if it were indicated that the workers had changed into protective clothing as advised. HERMES' rationale for rating the danger gauge at a crisis pitch, relatively safe, or somewhere in between, depends upon principles such as "poisonous substances are more dangerous than flammable substances," or "leaks are more dangerous as the flow rate increases (Clark et al., 1988)." Explosion and fire gauges consider such things as the flammability of identified substances, the location of ignition sources and dents in compressed gas tanks on sunny days, for example. Evacuation danger considers the minimum distance necessary to protect the public from flying objects, poisonous gases or other dangerous materials. HERMES ILLUSTRATIVE APPLICATION / 78 6.3.1.6. Tutorial system The tutorial system shows different situations to the trainee describing weather and other pertinent conditions. The trainee must analyze the scenario and evaluate explosion and fire potential as well as recommending an evacuation distance. Emergency response recommendations from various sources can also be posed to trainees, who must then justify their selection or rejection of each alternative based on the circumstances at hand. These recommendations are then compared to the selection HERMES would have made and the reasons why. 6.3.2. Satisfaction The project was deemed successful. Evaluations of three HERMES problem case-studies conducted by an independent expert produced "very satisfactory" results (Chang et al., undated). This was based on the observation that advice offered by HERMES was the same as the independent expert would have given. The general applicability of a system such as HERMES in small and large scale situations is appealing. It could be useful in a federal or provincial government agency, a municipality, or a private industrial site. Its rather generic decision-making framework is also very interesting from a planning point of view. The understood properties of hazardous materials remain the same, varying only in response to temperature or other situation-specific characteristics which must be input by the user, each case reflecting a different disaster scenario. The system might be plugged into action elsewhere in the country. As long as HERMES ILLUSTRATIVE APPLICATION / 79 environmental models appropriate to the various regions were part of the knowledge-base, and there were a sub-expert system capable of selecting the appropriate one, the advice offered by the system could reflect local knowledge and conditions. Alternatively, a system such as HERMES might be easily adapted by local planners elsewhere in the country, to incorporate local environmental models and a few local rules of thumb. 6.3.3. Future HERMES was developed as a proof-of-concept prototype. Many important features of expert systems can be time-consuming to develop and fine-tune. Before investing the resources and time in such endeavors, it was considered necessary to determine whether or not the problem-area of emergency response planning was suitable to the tool. Having established this fact to the satisfaction of all, plans to create a fully operational tool for emergency response planning and management, will require work in certain areas. The details in each area must now be resolved and fleshed out • knowledge base • uncertainty • models of environment • sense of time • explanation facility Improvements in these areas are not pipe dreams but quite attainable with the help of expert system technology. 6.3.3.1. Knowledge base The knowledge base must be expanded to cover as many hazardous materials and situations as possible. The amount of data necessary would be vast and a classification system has been suggested to represent materials according to their physical and HERMES ILLUSTRATIVE APPLICATION / 80 chemical characteristics (Clark et al., 1988). The possibility of interfacing with external data sources (e.g. municipal) is particularly exciting, as it would arm the system with accurate local city maps, soil and terrain maps or sewer systems. This would save considerable time in an emergency, and alleviate the pressure on a user who must otherwise try to obtain such information for decision-making. In terms of potential liability concerns, a system which is perceptibly transferable like HERMES, may ultimately require a caveat identifying those regions which its knowledge base may be ill-equipped to handle. 6.3.3.2. Uncertainty The issue of uncertainty is an important one to address and will be focussed upon during the current development phase. Currently, the prototype handles uncertainty in a very basic way in specific areas. With a general and more sophisticated scheme, it will be possible for the user to indicate the relative reliability of the information being entered into the system. This in turn would allow the system to indicate the relative reliability of each piece of advice it offers. 6.3.3.3. Environmental Models The models used to represent the environment must be improved. The current prototype only models how a substance might move through the air. There are several well-defined such models including substance absorption into soils and water which must be built into the system. The circumstances of a given accident would determine which model or combination of models is to be selected. An expert algorithm could be devised to select the most appropriate model. HERMES ILLUSTRATIVE APPLICATION / 81 The human understanding of the environment reflected in environmental models is always humble at best They may be better than nothing in accidents involving toxic materials, however. Doing further study or doing nothing at all because we admittedly do not fully understand the ecology of an area, are not realistic options when faced with a crisis. While these models may be useful, recognition of their limitations should perhaps be kept in mind at all times, as well as a willingness to review them as understanding grows. 6.3.3.4. Sense of time The system should be equipped to effectively schedule time-dependant resources (eg. firefighters) and to recommend additional equipment which would be valuable. Temporal reasoning would also allow the system to react appropriately once a user indicated that a leak had slowed or been patched (Chang et al., undated). If the system depended on a clock, it could monitor real-time events. The completed version of HERMES may be designed to handle temporal planning considered too complex for human experts. It is suggested that it may ultimately even devise previously unthought of but useful emergency response techniques. 6.3.3.5. Explanation facility An explanation facility for HERMES to justify and explain its solutions is crucial. Experts who disagree with a recommended course of action will not have faith in the system unless its reasoning is accessible and they can judge whether or not it is appropriate under the circumstances. HERMES ILLUSTRATIVE APPLICATION / 82 It is not considered a problem to develop a complete explanation system for the next phase of HERMES. A 'deep' knowledge domain is necessary, however, which is complex to develop and not considered worth it at the prototype level. 6.4. CHAPTER SUMMARY The Heuristic Emergency Response Management Expert System (HERMES) is intended to advise, assist and train emergency response staff in the management of hazardous materials accidents. The system was developed as a proof-of-concept prototype by the Alberta Research Council, Alberta Public Safety Services (APSS), and Emergency Preparedness Canada (EPC). On the basis of information supplied by the user, HERMES can estimate hazard levels e.g. fire and explosion risk; recommend procedures e.g. for evacuation; and offer important advice on the containment and clean up of the spill. HERMES operates on the same heuristic reasoning as human experts, but can perform algorithmic, mathematical computations when necessary, e.g. calculating the dispersal plume for a hazardous chemical. The work which HERMES is designed to do, namely, advise emergency response management, involves both immediate and post-crisis action. Immediate action involves the identification of the hazardous materials, the protection of human life, and the containment of the spill. Post-crisis concerns are clean-up procedures, evaluation and mitigation (if possible) of long term impacts on people and the environment Difficulties associated with the work include the overwhelming volume of information to HERMES ILLUSTRATIVE APPLICATION / 83 be searched and processed in times of stress; the multitude of possible chemicals; the uncertain and incomplete nature of information available for decision-making during a crisis; the impossibility of ensuring that the appropriate qualified expert is on duty when the nature and timing of accidents is inherently unpredictable; and the delays often experienced in locating someone with the necessary expertise during a crisis. The anticipated benefits seemed to outweigh the disadvantages of using an expert system in this case. Expert system features considered advantageous to this type of work include: the use of heuristic reasoning, the incorporation of uncertainty, the explanation facility, and the flexibility and adaptability of the system. Other appealing features are: the ability to preserve and distribute specialist knowledge, their training potential; consistency; efficiency; increased productivity; the ability to accumulate knowledge from several sources; and the possibility of re-using portions of the knowledge base for future systems. The limitations of expert systems which could cause concern in a problem area such as emergency response management include: the lack of creativity beyond that originally programmed; the inability to learn; the narrow focus; the lack of common sense; and the tendency not to know when its limits of knowledge have been reached. In light of these limitations, it is important to keep qualified staff to work with the expert systems. Staff must be alert to the weaknesses of the system and appreciative of the fact that it is merely a sophisticated tool. The problem area HERMES was designed to handle satisfies all the suitability guidelines outlined in Chapter 5. It relies heavily upon heuristic reasoning. The HERMES ILLUSTRATIVE APPLICATION / 84 advantages of using an expert system on the job outweighed the disadvantages, particularly if human experts are kept actively involved in the process. The knowledge domain for the work is highly specialized and focussed, and problems can usually be defined within reasonable boundaries. The complexity of the task is likely to fall within the fifteen minutes to several hours range much of the time (although not always). An enthusiastic and cooperative expert was available to participate in the system development Progress in the problem area can be improved in an incremental way, taking advantage of the effective system development method known as incremental prototyping. The fact that the problem area satisfied the task suitability guidelines for expert system development does not guarantee that the task is in fact suitable. These guidelines have yet to be tested extensively and refined because so few planning applications are in operation. The HERMES prototype has been judged highly successful by the development team, however. An independent expert ran three sample situations through the system and noted that the advice it gave was the same as a human expert would have given. HERMES has interesting implications to planning because it embodies a fairly generic problem solving strategy which might be applicable elsewhere in the country. Environmental models would have to be substituted for the different regions, however. It could be a useful system for both large and small organizations involved in emergency planning, public or private. Having been judged a success, the HERMES prototype is being extended into a fuller HERMES ILLUSTRATIVE APPLICATION / 85 operational system. The knowledge base, the uncertainty models, the sophistication of the environmental model, a sense of time and the explanation facility must be expanded and improved upon. The current knowledge base only handles a few chemicals, for example. The current environmental model only considers chemical movement through air rather than soil or water. HERMES would appear, to be an excellent example of an expert system application to emergency planning. The foreseeable advantages of a fully operational HERMES with improvements as suggested are clear. There is the possibility of saving lives and mitigating devastating environmental damage more than previously possible. HERMES offers the ability to manipulate mind-boggling amounts of information, to make necessary calculations much quicker than humans, and to interface with several information sources at once. It can offer advice based on the reasoning of the best human experts, but with the benefit of more information and in far less time than could be expected from human minds. In tandem, a human expert and an expert system may see truly remarkable improvements in emergency response work. The benefits of planning, well thought-out principles, collective experience, analysis of accidents and human response processes to date can be incorporated into HERMES; while the common sense, creativity and direct link to the circumstances at hand can come from the human users. C H A P T E R 7. S C R E E N E R I L L U S T R A T I V E A P P L I C A T I O N 7.1. S C R E E N E R - O R I E N T A T I O N 7.1.1. Developers SCREENER was developed by ESSA (Environmental and Social Systems Analysts Ltd.), a Vancouver based consulting firm. The client for this expert system was Transport Canada Airports Authorities Group. Funding was obtained through a research grant from the Department of Supply and Services, Canada. 7.1.2. Function The function of SCREENER is to operate as a decision making tool in the task of screening projects for environmental impact assessments. To lighten their load and allow for quick processing of non-threatening projects, SCREENER has been developed. SCREENER is not designed to make the final decision as far as implementation of any project is concerned, nor can it reach conclusions about the need for public review. It is designed to allow personnel with limited experience to make basic environmental impact decisions only. It ultimately recommends to the user whether the project can proceed with little or no modification, or whether it requires a more detailed assessment 86 SCREENER ILLUSTRATIVE APPLICATION / 87 7.1.3. Users SCREENER was originally developed for use by environmental officers of Transport Canada. These officers would have access to the detailed information pertaining to. each project which would be required by the SCREENER system. SCREENER is currently being adapted for use in similar situations in other organizations. As environmental concerns grow and there is increased pressure to account for environmental consequences of capital projects, many organizations are trying to internally screen their own projects. Tailored versions of SCREENER could be used by personnel who have a detailed knowledge of projects to be undertaken but who possess limited experience in environmental assessment After the initial screening (by SCREENER) only those projects determined to be of significant concern would then be contracted out to impact assessment professionals. There are versions of SCREENER underway for National Defense, Canadian Parks Service of Environment Canada, and the Coast Guard of Transport Canada. In each case, significant research must be done to develop a specific knowledge base (including environmental model) for the region in question. Other systems are also being contemplated in the future to be linked up with Geographic Information Systems. 7.1.4. Scope The full version of SCREENER as developed for the Transport Canada Airports Authorities Group was completed and ready for implementation in mid-1989. While it was not simply a prototype, the scope of the system was limited in SCREENER ILLUSTRATIVE APPLICATION / 88 recognition of the fact that environmental impact assessment cannot be easily codified to consider all possible factors, ecological, social or political. It was important that environmental officers be able to override the system's decisions if they considered that circumstances or additional information warranted such an action. The basis for such decisions would have to be explained and documented by the user. The Federal Environmental Assessment Review Office has identified nine decision codes to categorize potential outcomes of the Environmental Assessment Review process. The first six (0-5) can be established by SCREENER. Any projects of significant concern such as those rated as 4 or 5, are recommended for further review by more qualified people. Some of these may later be referred to public review or rejected outright if re-evaluated as 6, 7, 8, or 9, but SCREENER is not equipped to handle these ratings. 0 Initial assessment underway; no decision yet; 1 Automatic exclusion; project proceeds; 2 No significant adverse effects; project proceeds; 3 Potentially adverse effects may be mitigated with known technology; project proceeds; 4 Assess the proposal in greater detail (EEE required); impacts unknown; 5 Further study (EEE) required; ability to mitigate impacts is unknown -Significant impacts; 6 Refer proposal for public review by a panel - adverse impacts significant; 7 Refer proposal for public review by a panel - potentially significant adverse impacts; 8 Automatic referral for public review by a panel - potentially significant adverse impacts; SCREENER ILLUSTRATIVE APPLICATION / 89 9 Impacts unacceptable - either modify, rescreen or abandon project 7.1.5. Technical SCREENER was written in Arity/PROLOG version 5.1 and Microsoft C version 5.0 over a period of two years. The minimum equipment required to run the system would be an IBM AT (80286) or compatible microcomputer with a hard disk; 640K of conventional memory and 1 MegaByte of expanded memory. 7.2. P R O B L E M SUITABILITY 7.2.1. Problem Characteristics The problem at hand is to assist environmental officers of Transport Canada, Airports Authorities in preparing environmental impact statements. Federal policy requires that environmental impact statements be made for all capital projects in federal departments. Capital projects include a vast range of proposals from buying a desk, to building a runway. Efforts to question department staff on the detailed nature and difficulties of this job have been unsuccessful. The following interpretation of such details has been gleaned from interviews with the consultants who built SCREENER. The job of preparing environmental impact statements involves significant ecological expertise and knowledge of the land or region in question. Many projects submitted for assessment involve minimal impact if any at all, while others involve in-depth consideration and research. Public concern regarding environmental issues has increased SCREENER ILLUSTRATIVE APPLICATION / 90 and government departments are facing increased pressure to be accountable for its decisions in this regard. Problems associated with the work include an overwhelming workload; an inconsistency of assessment depending upon the individual reviewer and department; frustrating delays in project review; cost overruns due to environmental problems which should have been foreseen but were not; and poor documentation of decisions and the reasoning behind each. Rather than attempting to automate the entire assessment task, it was decided to develop a system which could automate the more easily processed assessments, and screen out those which were more complicated and had to be referred to qualified personnel for more detailed review. This would allow the experts to focus their energies and broad experience on the more significant projects. 7.2.2. Suitability Criteria The screening and basic assessment work done by SCREENER matches all the task suitability guidelines. Guideline (iii): a specialized knowledge domain and clear problem definition, is not met as clearly as others. Each guideline is discussed in the context of the case study following the summary in Table 9. SCREENER ILLUSTRATIVE APPLICATION / 91 Table 9. ENVIRONMENTAL IMPACT SCREENING MEASURED AGAINST TASK SUITABILITY CRITERIA Task S u i t a b i l i t y C r i t e r i a S a t i s f a c t i o n i ) Task invo l v e s h e u r i s t i c knowledge and cannot be handled by conventional systems. Y i i ) B e n e f i t s are considered worth the investment and c o s t s . I i i i ) S p e c i a l i z e d knowledge domain and c l e a r problem d e f i n i t i o n . m i v ) Task process not poorly understood. T v) True experts e x i s t and perform b e t t e r than novices using commons sense. Y v i ) Task takes 15 min. to several hours to perform (without expert system). T v i i ) A r t i c u l a t e expert i s w i l l i n g t o cooperate u n t i l system complete. Y v i i i ) P r o g r e s s can be made inc r e m e n t a l l y . T LEGEND: T • Yes, c r i t e r i a i s s a t i s f i e d . (Y)' - C r i t e r i a i s s a t i s f i e d , more of t e n than not. K « No, c r i t e r i a i s not s a t i s f i e d . SCREENER ILLUSTRATIVE APPLICATION / 92 (i) The task involves heuristic (non-numeric, non-algorithmic) knowledge and cannot therefore be satisfactorily handled by conventional computer programs. Yes, heuristic knowledge is involved. The developers of SCREENER claim that they would not even have tried to develop such a system were it not for expert system technology (Interview ESSA. July 1990). The declarative rule-based reasoning (approximately 700 rules) is also more efficient. Mathematical computations can be accommodated where desired. (ii) The benefits are worth the investment. Yes. In the clients' opinion, the anticipated benefits of using an expert system outweigh the potential disadvantages. In addition to those advantages itemized in Table 9, it was considered beneficial that expert systems could ensure public accountability by providing an automatic record of the decisions made; and reduce the potential for project delays and cost overruns due to unforeseen environmental concerns. (iii) The knowledge domain for the task at hand is specialized and narrowly focussed, and the problem is can be clearly defined. Moderately. Admittedly, the knowledge base for impact assessment in general is limitless, but for screening to the level of resolution which SCREENER attempts, the necessary knowledge base is reduced considerably. By allowing for recommendations of further study, the knowledge base is relieved of having to consider every conceivable situation, particularly in major capital projects or those which are unlikely to be undertaken at an airport The proposal to be evaluated is clearly specifiable and well-bounded. Those activities involved in constructing a building, or paving a runway, for example, are easy to define. The problem of what impacts it will have upon the site in question is less well bounded. The scope must be determined by the system based upon ecological SCREENER ILLUSTRATIVE APPLICATION / 93 relationships and patterns as perceived and declared by human experts. Project proposals which are not well bounded and clearly not well understood by the system will simply be referred for further impact assessment by human experts. The past experience and perception of experts whose knowledge has been used for an impact assessment expert system could prove to be a limiting factor. For this reason, the qualifications and reputation of the experts who create the knowledge base will necessarily be important for evaluating the system's performance. (iv) The task or decision-making process is not poorly understood. True. Humanity's deep understanding of ecosystem functioning is undeniably limited in spite of the fact that certain basic relationships have been perceived. Nonetheless, given that there are recognized disciplines of environmental planning and ecology, and an ever-growing consulting industry in doing environmental impact assessment work, the author assumes that there are accepted methods and decision-making processes to be followed in such work. There may be considerable debate concerning the appropriate methodologies. Impact assessment decision processes are also bound to change in the future as our comprehension and values change. From the environment's perspective, impact assessment decision processes may be poorly understood. But the consulting industry does have procedures. Even if flawed, they are presumably understood. The intent of this guideline is to ensure codability. For this reason, it has been rated as 'Y,' satisfying the guideline. Ethical issues such as whether the processes used by the industry at present are effective and deserve to be distributed through automation, must be dealt with by the profession and society. (v) True experts exist and do the job better than novices using common sense. SCREENER ILLUSTRATIVE APPLICATION / 94 Yes. True experts do exist There is much that experts do not know about the environment, but they generally know much more than an amateur. Recognized relationships in an ecosystem, specialized scientific knowledge, appreciation of different equilibria in the environment, and a healthy respect for that which is not known all form part of an expert's knowledge. They help the expert to estimate, what the impacts of a given capital project might be. (vi) The task at hand is generally complex enough to take more than fifteen minutes and less than several hours of an expert's time to perform. Yes. The screening of projects requires less time than an impact assessment itself. While months or years may be required for the impact assessment of some major capital project proposals, a screening to determine whether the project should be referred for a more detailed investigation does not take that long. With all the information at hand, the task would likely fall within this range. Those projects which are on an automatic exclusion list can be indicated as such at the beginning of the screening and it is not necessary to proceed through the more detailed procedure. (The automatic exclusion list, and the use of it for a given case would be documented however, for accountability reasons). Projects which are not automatically excluded but are of a moderate scope and can be assessed relatively easily as acceptable or not without further review are handled in that manner. (vii) An articulate expert is ready and willing to cooperate with the system developers until the system is complete. Yes. In this case, the expert and the system developers were the same. The firm which developed SCREENER (ESSA) is an environmental assessment consulting firm SCREENER ILLUSTRATIVE APPLICATION / 95 and had all the incentive necessary (profit and reputation) to cooperate fully in declaring, critiquing and effectively representing their screening process, (viii) Progress can be made incrementally. Yes. As different types of capital projects become common to the organization's activity, the knowledge base could be adapted. Changes to policies and laws can be incorporated as they occur. SCREENER ILLUSTRATIVE APPLICATION / % Table 10A POTENTIAL BENEFITS OF USING EXPERT SYSTEM FOR SCREENER PROBLEM (Main Benefits to Planners) MAIN EXPERT SYSTEM FEATURES OF INTEREST TO PLANNERS IS FEATURE BENEFICIAL FOR THIS PARTICULAR PROBLEM APPLICATION? 1. Expert systems can expand the range of plann i n g t a s k s to be automated. Tes. Models f o r t h i s type of work i n the past have been d i s a p p o i n t i n g . This task could not be e f f e c t i v e l y automated before. 2. Reasoning i s a c c e s s i b l e . Tes. The o f f i c e r doing the screening may wish t o q u e s t i o n or override a d v i c e . I t i s q u i t e conceivable that the user may know something the expert system does not. 3. Supports i t e r a t i v e and i n t e r a c t i v e problem s o l v i n g . Tes. Human understanding of environment impacts i s l i m i t e d and s t i l l growing. The system can be modified as understanding improves and values (what damage w i l l be to l e r a t e d ) change. 4. F l e x i b i l i t y t o adapt to changing c o n d i t i o n s . Tes. New p r o j e c t types may come in t o play at a i r p o r t s and system must be able t o evaluate the impact of t h i s new a c t i v i t y . S. I n c o r p o r a t i o n of h e u r i s t i c r e a s o n i n g . Tes. Experts have r u l e s of thumb, hunches, and may have i d e n t i f i e d i n e x p l i c a b l e p a t t e r n s i n the environment. Models of the environment must i n c o r p o r a t e these h e u r i s t i c s , as w e l l being a b l e t o perform mathematical computations where necessary. Does not assume q u a n t i t a t i v e p r e c i s i o n . =— Moderately. I n t e r i m a d v i c e may be requested based on incomplete in f o r m a t i o n , but a l l i n f o r m a t i o n about each p r o j e c t should . be a v a i l a b l e t o the o f f i c e r doing the e v a l u a t i o n . Environmental o f f i c e r s are not considered q u a l i f i e d and t h e r e f o r e not permitted t o g i v e confidence f a c t o r s t o t h e i r i n p u t , however. T h i s f e a t u r e would not be taken advantage of f o r t h i s j o b . _ SCREENER ILLUSTRATIVE APPLICATION / 97 Table 10B. POTENTIAL BENEFITS OF USING EXPERT SYSTEM FOR SCREENER PROBLEM SECONDART E.S. FEATURES OF INTEREST TO PLANNERS IS FEATURE BENEFICIAL FOR THIS PARTICULAR PROBLEM APPLICATION? 7. Preserve s p e c i a l i s t knowledge. Tes. Otherwise expensive c o n s u l t i n g knowledge may be used in-house to screen out those p r o j e c t s which are e a s i e s t t o review. 8. T r a i n i n g p o t e n t i a l . Tes. 9. Freeing experts f o r •ore c r e a t i v e work. Tes. True experts can work on those p r o j e c t s r e f e r r e d f o r more inrdepth assessments. They need not waste t h e i r expensive time on easy and t r i v i a l p r o j e c t s . 10. System not bewildered by vast amount of inf o r m a t i o n . Tes. This can save time and lead to more informed d e c i s i o n s . 11. I m p a r t i a l l y explore a l l hypotheses. -12. I n t e r a c t i o n a t each user's l e v e l . Tes. Users w i l l have varying degrees of f a m i l i a r i t y w i t h the subject knowledge. 13. Consistency. Tes. Evaluations must be done t o cons i s t e n t standards w i t h i n departments and the f e d e r a l government i t s e l f . 1*. Increased p r o d u c t i v i t y . Tes. Environmental o f f i c e r s at Transport Canada are c u r r e n t l y overwhelmed with assessments to do, ranging from buying a desk t o b u i l d i n g a c o n t r o l tower. 15. E f f i c i e n c y . •> Tes. R e l a t i v e l y small amount of input can generate w e l l - f o c u s s e d information needs. Asking user f o r missing information, system invokes only the r u l e s s p e c i f i c a l l y needed. 16. Reasoning easy to document. Tes. Can be u s e f u l f o r reviewing and e v a l u a t i n g a case afterwards i f necessary. 17. Runs en microcomputer. Tes. Each o f f i c e r eould be supplied with own computer. 18. E x p e r t i s e e a s i l y t r a n s f e r r e d . Tes. E x p e r t i s e needed in-house at a l l f e d e r a l a i r p o r t s i n country. 19. Experts can be inv o l v e d i n system development. Tes. 20. Accumulated knowledge from s e v e r a l sources. Tes. Oood idea to i n t e g r a t e knowledge from a number of people, each with t h e i r own area of e x p e r t i s e . 21. M o d i f i e d knowledge base u s e f u l f o r f u t u r e systems. Tes. — SCREENER ILLUSTRATIVE APPLICATION / 98 Table 11. DRAWBACKS OF USING EXPERT SYSTEM FOR SCREENER PROBLEM EXPERT SYSTEM LIMITATIONS OF CONCERN TO PLANNERS IS LIMITATION OF NOTABLE CONCERN FOR THIS PROBLEM APPLICATION? 1. U n i n s p i r e d . Not c r e a t i v e l i k e people. No. 2. • Cannot adapt, l e a r n without being t o l d . No. I t i s a l i m i t a t i o n t o be kept in mind but does not s e r i o u s l y threaten q u a l i t y of work. S t a f f w i l l have to keep system updated t o r e f l e c t c u r r e n t understanding of t h e i r l o c a l environment and v a l u e s . 3. Cannot i n t e r p r e t sensory data as w e l l - as people. Tes. T h i s can l i m i t the system's a p p r e c i a t i o n of the nature of the environment i t i s a s s e s s i n g impact upon. 4. Expert systems have narrow focus. Tes. V h i l e there are a l i m i t e d number of c a p i t a l p r o j e c t s l i k e l y at the A i r p o r t s A u t h o r i t y , the breadth of knowledge necessary may be wider than r e a l i z e d . T h i s i s why humans must be able t o o v e r r i d e the system's advice i f they f e e l s trongly about i t . The system i s only a t o o l . 5. Expert systems have no common sense beyond s p e c i f i c knowledge area. Tes t h i s c o u l d be a problem. Again, people must remain p a r t of the process and use t h e i r common sense i f i t seems necessary.. 6. Do aot know whea edges of knowledge reached. Tes. T h i s c o u l d be a problem because i t may assume i t knows a l l the p o t e n t i a l impacts f o r a g i v e n project when i t does not. 7. Cannot run autonomously f o r long p e r i o d s of time. No. An i n t e r a c t i v e system i s p r e f e r a b l e and people would have more confidence i n the r e s u l t s i f they were aware of i t s process. 7.3. PERFORMANCE SCREENER ILLUSTRATIVE APPLICATION / 99 7.3.1. Features This section highlights some of SCREENER's main features. The system is interactive and user-friendly. Users do not require much experience with the computer, although environmental knowledge is always an asset SCREENER improves the consistency of project assessments amongst different reviewers, and departments, if desired. The override feature may be used at any point within SCREENER. SCREENER will query the user for a reason to justify the override. This is printed on the final reports and a space is required for the user's signature. This feature allows for a certain amount of discretion in assessments, but discretion which is documented. An individual is ultimately accountable (at least on paper). The report automatically generated by SCREENER for each session (available in three formats) means that decisions and the reasoning for them are documented for all projects. A summary report, a detailed report, or a cross-impact matrix report can be sent to the printer. Examples of these are provided in the appendix of this thesis. The Summary Report lists the potential impacts for the major categories of environmental components (e.g. terrestrial animals, surface waters, employment/economy). The Detailed Report offers a complete description of each potential impact and description of mitigation measures. The Cross-Impact Matrix shows the expected impact for each environmental component described for the site. Records of all decisions are kept in a registry database accessible through dBASEIII+ software. SCREENER ILLUSTRATIVE APPLICATION / 100 SCREENER allows for re-screening. The user can store and retrieve a given project at a certain level of completion without having to start screening again from scratch. Some screenings can be quite long and complex. Project activities are grouped within project types, allowing the system to call up only information specific to those activities involved in that project type. This improves the efficiency of the system. 7.3.2. Satisfaction Comment has been unavailable from Transport Canada regarding satisfaction with the system. Whatever lessons there are to be learned may become obvious very soon, however. Meanwhile, the system developers have been satisfied enough to invest considerable time and energy in further developments of the system for other clients. In-house prototyping and development results are very encouraging. 7.3.3. Future An issue to be reckoned with in this area of application but which was not dealt with by SCREENER is that of cumulative impacts. Each capital project is assessed individually and there is no accomodation for the possibility that cumulatively, projects might exceed acceptable impacts on a given site. Future users of SCREENER will also have to ensure that the system is kept current with any changes in F E A R O policy and standards. SCREENER ILLUSTRATIVE APPLICATION / 101 7.4. CHAPTER SUMMARY SCREENER is an expert system developed by ESSA (Environmental and Social Systems Analysts Ltd.), for the Airports Authorities Group of Transport Canada. It is an expert system designed to screen capital projects and perform simple environmental impact assessments. It is intended to assist environmental officers of the Airports Authorities Group. The work of the environmental officers is affected by an excessive workload, a lack of consistency, project delays, and lack of effective documentation records for each project screened. Projects are extremely varied in nature and complexity. A significant amount of expertise is required to process some of the impact assessments, while others may fall onto automatic exclusion lists. The benefits of developing an expert system for this type of work include the following. Automating the task would alleviate the excessive burden on experts. Expert system technology could handle the heuristic knowledge used in this type of work. The accessible reasoning would allow users to judge and override advice if considered necessary. The decision-making process could be declared incrementally and gradually improved through prototyping. The system could be easily modified, reflecting new knowledge or project types. Additional advantages include the possibility of training less experienced officers; freeing the best experts for the more difficult cases; consistent assessments; accurate and complete records of each screening; accountability for decisions taken; the moderate cost of running the system on a microcomputer; the ease of transferring the expertise; and SCREENER ILLUSTRATIVE APPLICATION / 102 the possibility of accumulating knowledge from different sources. The limitations of expert systems which could influence effectiveness in this problem area include the following. Expert systems function best with narrow and highly specialized knowledge bases. Environmental assessment could become too broad. SCREENER has been limited to certain rating levels only, f o T this reason. Expert systems have no common sense, which is always a problem in any area. The system may not always know the limits of its own capability. An interactive process involving people is the best defense against these problems. Users must use their own common sense, query the machine when in doubt, and be aware of the system's limitations as a tool. It would not be regarded as infallible. The task which SCREENER is designed to accomplish meets all of the general suitability criteria suggested in Chapter 5.0. Guideline (iii) is met only partially, however, in the sense that the clarity of the problem to be tackled may be debated. There is a specialized knowledge domain and a description of the proposed activities is easily defined, but the possible environmental impacts may not all be known. The fact that the suitability guidelines are generally satisfied does not guarantee that the task is ideally suited to expert system use. They do indicate that according to current wisdom, this system has a good chance of being successful. For situations like this one in which manual methods and wider reaching policies are resulting in an overwhelming workload for staff, and unqualified people are making decisions poorly; the consistency and expert-level knowledge of this type of computer tool may prove to be beneficial. SCREENER ILLUSTRATIVE APPLICATION / 103 Feedback on its performance in the first few years may lend valuable insight into the use of expert systems in environmental assessment procedures and may even lead to changes in the suitability guidelines. SCREENER may not represent an improvement upon the environmental impact assessment process or screening process. It merely reflects the process that experts use and that which is acceptable to society and the consulting community at present That knowledge is distributed and made available to organizations trying to determine which projects can proceed with or without mitigation, and which projects must be referred for more in-depth assessments by better-qualified teams. While the final word is not in on SCREENER, it is an interesting example of expert system technology applied to an environmental planning task. Environmental planning as a field can be very broad but this smaller and more defined sub-task is potentially quite suitable to expert system use. CHAPTER 8. PLANCHECKER ILLUSTRATIVE APPLICATION 8.1. PLANCHECKER - ORIENTATION The name of this expert system has not yet been finalised but for simplicity sake will be referred to as PLANCHECKER. This was suggested by one of the system developers in response to my request for a name, but it may change before the system is completed and formally documented. 8.1.1. Developers P L A N C H E C K E R is a Canada Mortgage and Housing (CMHC) funded project in which a prototype expert system is being developed to check development plans for adherence to spatial separation requirements of Vancouver's building by-laws (Bluett, 1990). System design has been contracted to Dr. Edwin Bluett of E.B. Economics, Vancouver and Dr. Eric Heikkila, Assistant Professor of Urban and Regional Planning, University of Southern California. Vancouver's City Permits and Licenses Department is cooperating fully on the project in the role of client 8.1.2. Function The function of the expert system prototype, PLANCHECKER, will be to assist human Plan Checkers at City Hall in checking building plans for adherence to certain building by-laws. Plan Checkers must check building plans to ensure that they conform to the city by-laws. P L A N C H E C K E R processes information regarding a given development plan through its knowledge base and determines whether or not the plan satisfies spatial separation requirements. Spatial separation regulations are intended to prevent the spread of fire between buildings. A plan which does not satisfy these 104 P L A N C H E C K E R ILLUSTRATIVE APPLICATION / 105 requirements is not eligible to receive an R S - l / M building permit The task of plan checking for spatial separation, is one of ascertaining whether or not the number of openings (windows, doors, skylights, etc.) exceeds a critical value representing adequate separation from its neighbor and thereby ensuring greater fire safety (Bluett, 1990). This critical value depends upon the area of the building face, the distance of the building from the property line, and characteristics of building construction such as whether the building is sprinklered and whether the openings are glazed with regular glass, wired glass, glass blocks, or some other material (Bluett 1990). At present Plan Checkers use calculators and tables to help them with their work (Heikkila, Presentation, June 1990). P L A N C H E C K E R automates the process using the same tricks and calculations that the Plan Checkers currently employ. P L A N C H E C K E R also determines the appropriate by-law for judging spatial separation in a given case based on the knowledge base and information supplied by the user (Bluett, 1990). The knowledge for effectively evaluating spatial separation comes from building by-law articles, fire protection manuals and common procedures which are not itemized in the by-laws. Expert human Plan Checkers were interviewed extensively to determine their procedures, and decision making processes. The Plan Checkers' Manual was also used. 8.1.3. Users P L A N C H E C K E R is being designed for use by human plan checkers at City Hall. It is considered as an expert assistant and requires operators who are familiar with the work to a certain degree to feed it the appropriate information. P L A N C H E C K E R ILLUSTRATIVE APPLICATION / 106 The system should be of most use to novices who are still learning and who will lean significantly on its advice. It is claimed that experienced plan checkers should not be held back by the system, and that they may find it useful for catching errors which are easy to make in a tedious and repetitive and complex job. 8.1.4. Scope P L A N C H E C K E R is a prototype being undertaken to determine the feasibility of an expert assistant for the Plan Checkers at City Hall. Its scope has been reduced to checking plans for spatial separation, but a final and fully operational system, if developed would cover a broader range of plan checking tasks including exiting and yard setback analysis. Spatial separation verification can take from 20% to 50% of a Plan Checkers time. Development of the P L A N C H E C K E R prototype began in April of 1990 and is expected to be complete by the end of August, 1990. 8.1.5. Technical P L A N C H E C K E R has been developed on the VPExpert Shell. This expert system building tool was selected because it was relatively cheap, it offered the type of reasoning considered necessary for the task at hand, and it would allow for a much faster development of the prototype than would be possible if the program were to be written in Lisp or Prolog. To date, the program developers are very happy with the performance of the shell (Interview. Bluett and Heikkila, 1990). The final P L A N C H E C K E R can run on an X T computer. Computers of more power P L A N C H E C K E R ILLUSTRATIVE APPLICATION / 107 allow the system to run much faster. 8.2. P R O B L E M SUITABILITY 8.2.1. Problem Characteristics 8.2.1.1. Description The Plan Checker's job is extremely complex. There are a myriad of rules used to evaluate plans and it takes approximately one year for an employee to become proficient in the job. The following description is a simplified, skeletal version of the process. There is a rough rule of thumb and there is a more accurate assessment process called ratioing to determine whether or not the building plan passes the spatial separation requirements (Bluett, 1990). If the plan satisfies the rough test, the building passes for sure. If it fails that test, the more accurate ratioing process must be tried. The rough test compares the sum total of the areas of all the openings to a limiting value based on the distance of the nearest part of the exposing building face to the property line. The ratioing test, on the other hand, considers each plane separately if they are at different distances from the property line. The amount of openings permitted on a plane would increase, the further was a plane from the property line. Many regulations have been drawn up to account for the increased safety represented by openings on planes further back from the property line. Openings on each plane P L A N C H E C K E R ILLUSTRATIVE APPLICATION / 108 are projected to a certain distance depending on the juxtaposition of other openings on that plane, flanking walls and distance of openings on flanking walls from the property lines (Bluett, 1990). The relationship of these objects must also be checked against rules for establishing the limiting distance for each opening. The ratio of the minimum limiting distance for the building, to the limiting distance of the opening, is used to reduce the area of each opening for calculation purposes. The reduced area of the opening is then used in the final spatial separation calculation, reflecting its greater safety factor from being further from the property line and having certain relationships to other characteristics of the building face. 8.2.1.2. Difficulties Associated with the Work The following list itemizes some of the problems encountered in the present system of plan checking (Phone interviews with management of Permits and Licensing Department, August 1990). * The work is complicated and employees must gain significant experience to become proficient The job does not require a high level of formal education but is a specialised form of expertise requiring specific in-house training. * There is high turnover on the job (average six months), and new, inexperienced personnel are incapable of delivering the quality of service the department would like to provide the public on a regular basis. * Staff consider the rank and salary scale of the job to be underrated, and the experienced and bright personnel are constantly leaving for jobs of higher rank and salary. * The job is tedious and repetitive. PLANCHECKER ILLUSTRATIVE APPLICATION / 109 It is a high-pressure job. Angry developers are always pushing for faster and more lenient processing. Minimal job satisfaction. Inconsistent interpretation of the rules by different Plan Checkers. Developers seek out checkers known to be lenient There is no incentive to do good work. Union pay scales mean that salary increases are based on time of service rather than productivity or error rates. Mistakes can be serious. Inappropriately evaluated plans can cause anger and legal problems, as well as costing the plan proponent and the City of Vancouver time and money. 8.2.2. Suitability Criteria The task which P L A N C H E C K E R is designed to perform meets all of the suitability criteria suggested for task evaluation. Each guideline is summarized in Table 12 and then discussed. Table 12. P L A N CHECKING MEASURED AGAINST TASK SUITABILITY CRITERIA Task S u i t a b i l i t y C r i t e r i a S a t i s f a c t i o n i ) Task i n v o l v e s h e u r i s t i c knowledge and cannot be handled by conventional T systems. i i ) B e n e f i t s are considered worth the investment and c o s t s . i i i ) S p e c i a l i z e d knowledge domain and c l e a r problem d e f i n i t i o n . i v ) Task process not poorly understood. v) True experts e x i s t and perform b e t t e r than- novices using commons sense. v i ) Task takes 15 min. to several hours to perform (without expert system). v i i ) A r t i c u l a t e expert i s w i l l i n g to cooperate u n t i l system complete. • v i i i ) P r o g r e s s can be made incrementally. L E G E K D ; T • Yes, c r i t e r i a i s s a t i s f i e d . (Y) * C r i t e r i a i s s a t i s f i e d , more o f t e n than not. N « No. c r i t e r i a i s not s a t i s f i e d . P L A N C H E C K E R ILLUSTRATIVE APPLICATION / 110 (i) The task involves heuristics (non- numeric, non- algorithmic) and cannot therefore be satisfactorily handled by conventional computer technology. Yes. There are heuristic shortcuts taken by Plan Checkers. Consultation with human Plan Checkers provided the system designers with equivalences to the by-laws that are routinely used in assessing building plans for spatial separation, what tests they perform and calculations they make. The by-laws and fire regulations are encoded into the knowledge base, but equally valuable is the expert knowledge gathered directly from the Plan Checkers themselves. Plan checkers would not bother doing certain calculations, for example, if a quick ruleof-thumb test assured them that it wasn't worth their time to get into more specifics (see 8.2.1.1 - Description). (ii) The benefits are worth the investment Yes. The benefits anticipated from an expert system tool in this case are outlined in Table 13 A and 13 B. Of particular interest to the client is the increased work quality and productivity potential; the increased consistency of rules interpretation; and the retention of undocumented rules-or-thumb which are lost as the experienced plan checkers leave. The limitations are outlined in Table 14. There were none of major concern. There may be negative social consequences to the use of P L A N C H E C K E R which will not be recognized until a future date. The workload expected of each plan checker is bound to increase and, if there is not enough work to keep all busy at the accelerated pace of production, some may be laid off. It is not inconceivable that computers with later versions of P L A N C H E C K E R may even be put on the public counter in the Planning Department at City Hall some day, allowing the public to check their own plans, and receive automatic approval in one session. P L A N C H E C K E R ILLUSTRATIVE APPLICATION / 111 From the department's point of view, the workload is so excessive at present that they deny the possibility of any layoffs. Plan checkers themselves seem keen to use this tool to make their job easier. There was no resistance to cooperating with the consultants for knowledge declaration purposes (Phone interviews with Management, Permits and Licensing Dept., Aug. 1990). (iii) The knowledge domain for the task is specialized and narrowly focussed and the problem can be clearly defined. Yes, the knowledge domain for plan checking is specialized. This is why the high staff turnover mentioned in the preceding section is a problem. The problem can also be specified within clearly defined terms. (iv) The task or decision making process is not poorly understood. The task is relatively complex but is clearly understood, at least by those who developed the process or have worked with it (v) True 'experts' exist, meaning it is acknowledged that they can do the job better than novices. Yes. True experts in this problem domain do exist While the problem domain may not be difficult for the average person to eventually learn, it is difficult for the average person to accomplish the task without adequate instruction or training. There is no doubt that experienced plan checkers (experts) can do the job better than novices. There is also a generally agreed upon approach which is supported by the Plancheckers Manual. (vi) The task at hand is generally complex enough to take more than fifteen minutes and less than several hours of an expert's time to perform. Yes. The average screening takes two and a half hours. P L A N C H E C K E R ILLUSTRATIVE APPLICATION / 112 (vii) An articulate expert is ready and willing to cooperate with the system developers until the system is complete. Yes. P L A N C H E C K E R system designers have enjoyed the full cooperation of some of the most experienced Plan Checkers in the City Permits and Licenses Department Much of the Plan Checkers knowledge has been declared in the Plan Checkers Manual. Revisions, modifications, additional tricks of the trade and comments have been offered by these employees in interviews and working sessions with the system developers. They have also been involved in the critiquing of the system. During preliminary runs of the system, Plan Checkers picked out reasoning errors and introduced shortcuts because the system "wasted their time," taking them through things step by step which they had already ruled out as impossibilities in their heads. Similar observations were made in the latest work session, August 9, 1990, but the system appears to be nearly complete. (viii) Progress can be made incrementally. Yes. A system such as this can be developed incrementally. An evolutionary prototyping process has been used quite effectively to bring this system to near completion in a matter of months. Once complete and functional, further modifications to the by-laws, fire-regulations, or even conventions for evaluating a plan could easily be accommodated through the changing of rules in the knowledge base. (This assumes that the modifications do not involve a different logic than the inference engine supports.) PLANCHECKER ILLUSTRATIVE APPLICATION / 113 Table 13 A POTENTIAL BENEFITS OF USING EXPERT SYSTEM FOR PLANCHECKER PROBLEM (Main Benefits To Planners) MAIN EXPERT SYSTEM FEATURES OF INTEREST TO PLANNERS IS FEATURE BENEFICIAL FOR THIS PARTICULAR PROBLEM APPLICATION? ! 1. Expert systems can expand the range of pla n n i n g t a t k a t e be automated. Yes. This task c o u l d not be e f f e c t i v e l y automated b e f o r e . 2. Reaeoning i s a c c e s s i b l e . Yes. T h i s feature helps t r a i n inexperienced plancheckers and a l s o helps them o f f e r a d v i c e t o developers as to what improvements to the plan are necessary. 3. Supports i t e r a t i v e and i n t e r a c t i v e problem s o l v i n g . Yes. The system's knowledge base can be modified as understanding improves. The WHAT IF f u n c t i o n a l s o allows the Plan Checker t o t r y d i f f e r e n t options t o f i g u r e out what changes might a l l o w the plan to pass. T h i s can be h e l p f u l advice f o r the developer. 4. F l e x i b i l i t y t o adapt t o changing c o n d i t i o n s . Yes. By-lavs can change. One changed while the prototype was being developed t h i s summer. The p r i n c i p l e s of thermodynamics which are the b a s i s f o r many of the r u l e s don't change though. 5. I n c o r p o r a t i o n of h e u r i s t i c r easoning. • Yes. Does not assume q u a n t i t a t i v e p r e c i s i o n . No. This system does not take advantage of t h i s f e a t u r e . Plans are rated e i t h e r as Pass or F a i l System needs t o know c e r t a i n dimensions p r e c i s e l y i n order t o guarantee a pass. Plans can always be measured and s c a l e d so incomplete inforsxation should never be a problem. PLANCHECKER ILLUSTRATIVE APPLICATION / 114 Table 13 B. POTENTIAL BENEFITS OF USING EXPERT SYSTEM FOR PLANCHECKER SECONDARY E.S. FEATURES OF INTEREST TO PLANNERS IS FEATURE BENEFICIAL FOR THIS PARTICULAR PROBLEM APPLICATION? 7. Preserve s p e c i a l i s t knowledge. Yes. High turnover means people with a good grasp of a l l the t r i c k s of the trade are r a r e . S. T r a i n i n g p o t e n t i a l . Yes. This should improve the q u a l i t y of work that inexperienced people do as w e l l as p r o v i d i n g c o n s i s t e n t reasoning on demand. 9. Freeing experts t o r •ore c r e a t i v e work. No. This i s not r e a l l y an issue f o r t h i s job. I t can r e l i e v e undue workload from the few experts who know what they are doing, though. 10. System not bewildered by vast amount of in f o r m a t i o n . Yes. This can save time and reduce e r r o r s on p a r t i c u l a r l y complicated cases. 11. I m p a r t i a l l y explore a l l hypotheses. Not r e a l l y an i s s u e . 12. I n t e r a c t i o n at each user's l e v e l . Yes. Experts have v a r y i n g degrees of f a m i l i a r i t y of d i f f e r e n t s i t u a t i o n s . 13. Consistency. Yes. Inconsistent i n t e r p r e t a t i o n of the r u l e s i s a s e r i o u s problem i n the department. • 14. Increased p r o d u c t i v i t y . Yes. PLANCHECKER i s expected t o increase p r o d u c t i v i t y and reduce the backlog of permit a p p l i c a t i o n s . IS. I f f i c i e n c y . Yes. R e l a t i v e l y small amount of input can generate w e l l - f o c u s s e d information needs. Asking user f o r missing information, system invokes only the r u l e s s p e c i f i c a l l y needed. 16. Reasoning easy to document. Yes. Can be u s e f u l i f f i l e must be re-opened f o r a r e - e v a l u a t i o n or f o r a renovation or b u i l d i n g a d d i t i o n . At present, plancheckers have t r o u b l e understanding each others notes i n these s i t u a t i o n s . 17. Runs en microcomputer. Yes. I t becomes more a f f o r d a b l e and each plancheeker may get t h e i r own computer to work w i t h . IB. e x p e r t i s e e a s i l y t r a n s f e r r e d . Yes. 19. Experts can be in v o l v e d i n system development. Yes. This has been taken advantage o f . 20. Accumulated knowledge from s e v e r a l sources. Yes. Several plan checkers have p a r t i c i p a t e d i n the work ses s i o n s and debate i n t e r p r e t a t i o n s u n t i l they a l l s e t t l e on an agreed upon implementation of the r u l e . 21. M o d i f i e d knowledge base u s e f u l f o r future systems. Yes. P L A N C H E C K E R ILLUSTRATIVE APPLICATION / 115 Table 14. POTENTIAL DRAWBACKS OF USING EXPERT SYSTEM FOR PLANCHECKER PROBLEM EXPERT SYSTEM LIMITATIONS OP CONCERN TO PLANNERS IS LIMITATION OP NOTABLE CONCERN rOR THIS PROBLEM APPLICATION? 1. U n i n s p i r e d . Not c r e a t i v e l i k e people. No. Th i s i s not a c r e a t i v e job and the absence of c r e a t i v i t y i n the t o o l i s not a l i m i t a t i o n . 2. Cannot adapt, l e a r n without being t o l d . Moderate on l y . When by-laws change, m o d i f i c a t i o n s can be made. Not a s e r i o u s problem. 3. Cannot i n t e r p r e t sensory data as w a l l as people. No. Th i s s k i l l i s not necessary for the job. 4. Expert systems hava narrow l o c u s . No. T h i s task i s so s p e c i a l i z e d that i n f o r m a t i o n i s not r e a l l y r e q u i r e d from any other f i e l d . 5. Expert systoas have no common sense beyond s p e c i f i c knowledge araa. Yes t h i s c o u l d be a problem. Dimension e r r o r s on plans might be detected as n o n s e n s i c a l by humans but w i l l not be by the computer. Plancheckers should p i c k these up as they enter dimensions i n t o the system. Data-entry e r r o r s c o u l d a l s o cause problems i n t h i s r e s p e c t . 4. Do not know whan edges of knowledge reached. No. T h i s i s a f a i r l y l i m i t e d domain and t h i s would not be a problem. 7. Cannot run autonomously f o r long p e r i o d s of time. No. The system only performs 20% to 50% of a plan checker's job so they are not expected t o run autonomously. P L A N C H E C K E R ILLUSTRATIVE APPLICATION / 116 8.3. PERFORMANCE 8.3.1. Main Features Only the keyboard can be used to input information when requested by the computer. A mouse and graphic features have not been used. The W H A T IF feature of the program allows the user to experiment and see how the allowable unprotected opening percentage changes. By pressing WHAT T F (after receiving a calculated percentage) the arrow keys can be used to select different variables, eg., "Is the building sprinklered?" As the answers regarding these variables are changed, the allowable openings percentage is recalculated with the new information. This percentage is itself a key variable in testing both the crude rule for spatial separation and for checking spatial separation using ratioing techniques. There is the potential to read or write to data bases or spreadsheets if City Hall want this feature, although it has not yet been developed on the prototype. 8.3.2. Satisfaction Both client and developer satisfaction cannot be determined at this point because the prototype is incomplete. There were 'bugs' to be worked out at the last presentation and ideas and comments from the Plan Checkers still to be incorporated. Both parties appear enthusiastic about the potential for the program, however. The system developers have also expressed satisfaction with the performance of the shell used to build P L A N C H E C K E R . P L A N C H E C K E R ILLUSTRATIVE APPLICATION / 117 8.3.3. F U T U R E The city plans to use the completed P L A N C H E C K E R as soon as it is finished. If the expected benefits materialize, the City may press for financing to develop exiting and yard setback prototypes to aid the plan checkers in other portions of their job. 8.4. CHAPTER SUMMARY P L A N C H E C K E R is a C M H C funded expert system intended to assist human plan checkers in verifying spatial separation requirements for building plans. The City of Vancouver's Permits and Licensing Department is the client Dr. Bluett of E.B. Economics and Dr. Heikkila of University of Southern California are developing the system. Spatial separation is important for fire protection reasons and if a plan fails to pass these requirements, it is denied a building permit The planchecker must determine whether the number of openings on the building (eg. doors and windows), exceeds a critical value representing adequate separation from its neighbor. PLANCHECKER operates using the same tricks and calculations as the human plancheckers. Rules of thumb and quick tests are an integral part of the process. The benefits of P L A N C H E C K E R anticipated by the client include increased accuracy and speed of plan checking, the retention of good plan checkers' knowledge and techniques in the face of high job turnover, and a facility for training new staff. There are no anticipated limitations of notable concern. The absence of common sense is a minor concern but human personnel must work with the system and must counteract this weakness. P L A N C H E C K E R ILLUSTRATIVE APPLICATION / 118 The task which P L A N C H E C K E R is designed to perform meets all the suitability guidelines suggested in chapter 5.0. It involves heuristic reasoning and cannot be accommodated by conventional computer programming. The benefits outweigh the disadvantages in the view of the client and system developer. The knowledge domain is highly specialized and the problem can be clearly defined. True experts exist and can perform the job infinitely better than the average person with no training. The task takes approximately two to three hours to perform on average. Articulate and enthusiastic experts have been willing to cooperate and critique the system throughout the development phase. Progress has been made incrementally and may be improved in this manner in the future. This particular expert system application is still in progress and one must be careful not to draw conclusions prematurely. It would appear at this stage, however, that the task is a suitable one for expert system development The automation process is progressing rather smoothly and with remarkable speed, using the expert system technology. Most of the advantages expected from the system still seem likely. CHAPTER 9. CONCLUSION 9.1. S U M M A R Y 9.1.1. Objective The objective of this work was to assess the fit between planning tasks and expert system technology. To do this it was necessary to look at the nature of expert systems and the nature of planning decision-making. The features which distinguish expert systems from other computer tools were discussed, as were the task characteristics required for the tool to function reliably. Three Canadian systems now at the forefront of expert systems application to planning have been used as illustrative applications. 9.1.2. Description Expert systems are a form of artificial intelligence and represent a significantly different class of software than conventional computer software to date. They are distinguished by their ability to mimic human decision-making and to use judgmental, non-numeric problem solving methods. An expert system is basically comprised of a reasoning component (inference engine) and a separate knowledge component (knowledge base). The user-interface permits a user to interactively consult with the system, much as they would consult a human expert for help in a particular problem domain. The transparency of an expert system's reasoning and the ease with which the system can be modified are additional distinguishing characteristics. 119 CONCLUSION / 120 9.1.3. Advantages It has been suggested that expert systems should appeal to planners because they promise to expand the range of planning work which can be addressed by computers. The system's reasoning is accessible and visible to the user. The iterative and interactive process of defining problems and goals, the flexibility to deal with constant change, the ability to deal with uncertainty or incomplete information should also appeal to planners. Expert systems can incorporate the judgmental and non-numeric aspects of decision-making which cannot be handled by conventional computer programs. The nature of many planning tasks can benefit from this feature. Expert systems also have potential as training aids for staff, can preserve and distribute critical knowledge, can foster consistency and efficiency of decision-making, can offer an explanation for all decisions taken, and can maintain a record of all consultations for later study or accountability. 9.1.4. Disadvantages There are limitations to expert systems of which the planner should be aware. Expert systems cannot be as creative as people. While expert systems may be faster and capable of processing much more information than people, the decision-making processes which are followed can be no better than were those of the human experts. Expert systems cannot learn new reasoning, and cannot interpret sensory data directly the way people can. An expert system does not always know when a problem is beyond the limits of its knowledge base, and cannot function unattended. Expert systems cannot employ common sense, and have trouble working with a broad focus -CONCLUSION / 121 again something of which people are more capable. The declaration of knowledge for the knowledge base can be a difficult process and human experts can be distracted for long periods of time as they participate in the building of a system. The benefits must be considered worth the significant investment of time and money resources. Other potential problem areas of which to be wary include the planner's ability to recognize when a planning task is suitable to expert system technology and when it is not; the possibility of liability for advice given on the system or references used; and the organizational changes which the introduction of this new technology may bring to planning offices, and possibly society. Potential social impacts such as layoffs, less performance differential between experienced and inexperienced personnel in specialised domains, and less opportunity for staff to use discretion in dealing with each case, could prove to be disruptive. 9.1.5. Suitability Guidelines Expert system literature suggests that unless a task meets certain general criteria, it may be difficult to code effectively and may therefore compromise the system's performance of that task. There are also many advantages and disadvantages associated with the use of expert systems, which should be considered in the task evaluation process. Eight guidelines have been identified so that planners can assess a given task's suitability for expert system use. A suitable task would: involve heuristic (non-numeric) knowledge; be able to ultimately benefit from expert system technology, given the advantages and disadvantages; be within a specialized domain and have a clearly defined problem statement; not have a CONCLUSION / 122 poorly understood decision-making process; require solution by an expert, because a layperson would be incapable; take between fifteen minutes and several hours to perform; have the benefit of an available, articulate and cooperative human expert for the expert system building process. It should also be possible to progressively improve upon the declaration of how to solve the problem (task) at hand. Planning, in general, does not meet all of these criteria but there may be a great number of planning functions and sub-tasks which do. It is important to evaluate the suitability of a given task before investing resources in the development of an expert system. Planners may wish to ask themselves whether it is necessary for a task to meet all criteria in order to be considered suitable. A reasonable match may be more practical. A task which does not squarely meet all of the criteria ideally required for optimal performance of an expert system, may meet even fewer criteria necessary for other computer models or manual methods. In many cases, expert system technology may prove to be an imperfect but preferable tool compared to the alternatives. Three Canadian expert system applications namely, HERMES, SCREENER, and P L A N C H E C K E R have been discussed to illustrate the potential compatibility of planning and expert systems. These were the most complete examples of planning applications found and two of these are still in the prototype phase. The development of planning applications of expert systems in this country is in its infancy. CONCLUSION / 123 9.1.6. Hermes HERMES (Heuristic Emergency Response Management Expert System) is currently being developed by the Alberta Research Council, Alberta Public Safety Services, and Emergency Preparedness Canada. HERMES was developed as an intelligent assistant and tutor for emergency response management personnel. On the basis of user input, the system offers advice for the protection of human life and the environment, the containment and clean-up of hazardous materials, and for mitigation measures to prevent avoidable long term damage to people and the environment A successful proof-of-concept prototype has been completed and the project is now being expanded to produce a fully operational system. The work which HERMES is expected to perform satisfies all but one of the task suitability criteria. Criteria (vi), that tasks must take between fifteen minutes and several hours to solve was rated as partially satisfied because some tasks may be out of this range. It is anticipated that emergency response in the event of a disaster will be much better with a system like HERMES, than is currently possible. The flexibility of the system, the incorporation of heuristic reasoning, the ability to work with uncertain and incomplete information, and the accessibility of the reasoning hold significant appeal. The anticipated benefits include swifter and more accurate diagnosis of situations, and better advice, particularly when a particular type of expert is unavailable. Ultimately this should translate to more lives saved and less damage sustained to the environment CONCLUSION / 124 Expert system limitations which could cause concern in the HERMES case study include the fact that expert systems do not always know when they are at the limits of their understanding of a problem; they lack common sense; and cannot be more creative than the experts who programmed them. Emergency management personnel would have to use HERMES as no more than a tool, to be combined with their own talents and skills. The general applicability of HERMES to small and large scale situations is appealing. The rather generic decision-making framework which it represents is also interesting from a planning point of view. The situation specific details can be input and processed at the time of the disaster. The anticipated benefits and the success of the prototype at this point would suggest that this is an excellent example of expert system technology applied to emergency planning. 9.1.7. Screener SCREENER was developed by Environmental and Social Systems Analysts Ltd. of Vancouver for the Airport Authorities Group of Transport Canada. SCREENER is intended as a decision-making tool in the screening of projects for environmental impact assessment purposes. SCREENER allows the environmental officers of the Airports Authorities Group to quickly screen and sort the overwhelming number of capital projects (ranging from the purchase of a new desk to building a runway). SCREENER could be used by less experienced personnel to process impact statements for simpler projects and to screen out those requiring referral to more qualified personnel. CONCLUSION / 125 SCREENER does not make the final decision regarding approval of any projects. Detailed accounts for each screening are kept by the system and printed reports are available after each session in the form of a summary, a detailed report, or a cross-matrix chart The work which SCREENER performs meets all but one of the expert system task suitability criteria. It only meets guideline (i) partially. The knowledge domain is specialized but not as narrowly focussed as would be ideal for an expert system. The scope of the system has been limited to a certain sub-task of environmental impact assessment for this reason. Main expert system features of interest to planners in this case-study would be the flexibility to adapt to changing values and understanding of environmental impact assessment; and the accessible reasoning, which allows the environmental officer to override the system's advice if s(he) considers it inappropriate. More consistent screening and assessment standards, greater accountability for decisions taken, and increased efficiency are among the anticipated benefits from Transport Canada's viewpoint Expert system limitations which could negatively influence this type of an application include a lack of common-sense and a lack of appreciation that the outer limits of knowledge have sometimes been reached on a problem. SCREENER is a fully operational system and while feedback has not been obtained from the Airports Authorities Group for this thesis, the system developers seem pleased CONCLUSION / 126 with the results so far. The system was implemented in 1989. The perceived success of SCREENER has prompted contracts for modified versions of the system for National Defense, Canadian Parks Service of Environment Canada and the Coast Guard of Transport Canada. The final results are not yet available but SCREENER is an interesting application of expert systems technology to environmental planning. SCREENER may or may not represent an improvement upon screening or initial impact assessment techniques, but it does represent an efficient, consistent, and flexible tool for performing work to the contemporary standards of this established environmental consulting firm. Environmental planning is a broad field, but this smaller and more modest subtask of screening projects seems in this instance to be quite compatible to expert system use. It will be interesting to note whether feedback from the Airports Authorities Group in the next few years affirms this. 9.1.8. Planchecker P L A N C H E C K E R is a prototype expert system being developed by E B . Economics for the City of Vancouver. The project has been funded by C M H C and the intent is to determine the feasibility of developing an intelligent assistant for Plan Checkers at City Hall. Building plans must be checked to ensure their conformity to spatial separation requirements in order that they may be granted development permits. P L A N C H E C K E R is the least evolved of the three applications (development commenced Spring 1990). The task it is designed to perform meets all of the expert system task suitability criteria and developers are enthusiastic about the potential success of this CONCLUSION / 127 application. Development is progressing very quickly and without major difficulty. Expert system features which can be of particular benefit in this type of application include: the incorporation of heuristics, e.g. plan checkers' rough tests and shortcut reasoning; the accessibility of reasoning (to explain why a plan does not pass); the iterative problem solving, eg. WHAT IF variable x were changed, would the plan then pass the spatial separation requirements?. The client is anticipating that P L A N C H E C K E R will help staff to achieve a much quicker, consistent and accurate level of service then is currently possible. The job is complex and specialized but there is a high employee turnover. It is hoped that P L A N C H E C K E R will also prove to be a good training tool. Performance limitations generally associated with expert systems are unlikely to cause concern in this type of application. The only weakness might be that a lack of common sense would prevent the system from detecting data-entry errors ('typos') as the human planchecker entered information. Hypothetical negative consequences of P L A N C H E C K E R from a sociological point of view might include an increased workload for some Plan Checkers and job loss for others. The human and social consequences of applying expert system technology may generate more debate in a non-critical domain such as this, than in life or environment saving cases such as HERMES and SCREENER. It should be noted, however, that management claim layoffs are unlikely given the workload and that employees seem eager to take advantage of the new tool. CONCLUSION / 128 9.2. CONCLUDING C O M M E N T Although expert system capabilities have been gaining the attention of planners in the last two to three years, the development of fully operational planning applications is still in the exploration phase. General observations concerning the three Canadian applications reviewed are as follows. All three problem areas satisfied the task suitability guidelines to a reasonable degree. Most guidelines focussed on the 'codability' of the task but there was also emphasis on weighing the advantages and disadvantages. The advantages seemed to outweigh the disadvantages both from the developer and client's perspective in each application. Each of those expert system features which were suggested as main points of appeal to planners was exploited by one or another of the sample applications. The HERMES problem area could benefit from all these. The SCREENER and P L A N C H E C K E R problems could benefit from all, except the ability to handle uncertain and incomplete information. Each application could also benefit from several of the additional features of appeal which were suggested. Finally, each application seems, at this admittedly early stage, to be enjoying success. While these problem areas have met the suitability guidelines, and do appear to have supported successful expert system applications, the guidelines are not a guarantee of success. The development and validation of guidelines to assess a planning task's suitability to the technology is still in its infancy. The suitability guidelines outlined in this thesis have set reasonable boundaries for expert system application to planning tasks. The outer boundaries remain undefined because the full scope and power of the CONCLUSION / 129 tool is still relatively untested in the planning field. Some readers may feel that planning has been inadequately addressed through the illustrative applications. The problem areas are rather technical sub-tasks within planning, even though they do involve heuristic and judgemental knowledge. The three applications reviewed did not sretch the guidelines the way some planning tasks might Examples centering on land-use control, design, or public participation issues may have exposed some different and interesting opportunities and liitations. It must be remembered that there are currently few planning applications from which to choose; it can be difficult to obtain programs being developed in different cities and countries; and it can be difficult to find adequate documentation or written accounts of systems still under development All these factors affected the range of applications used for illustration purposes. Most expert systems related literature and experience centers on the clearer and more tested waters of medicine, engineering, and even law, rather than planning. It may be premature, therefore, to draw any definitive conclusions about the compatibility of planning work and expert system technology. Time and experience will bring more specific lessons and wisdom in terms of the optimal relationship, than are possible now. Mistakes and successes will refine theory's claims and expectations. It is not premature, however, to reach a tentative conclusion. Planners must start seriously considering the significance of expert system technology and its relation to their field. With a significant number of prototypes currently under development and CONCLUSION / 130 operational systems just beginning to appear, planning professionals will have no choice but to respond to the advent of the technology in one way or another. The literature review indicated that there might be a valuable role for expert system technology in planning. Criteria for selecting those tasks most suitable to expert systems application were suggested and numerous planning applications have been initiated. At this point on the learning curve, the experiences of the three Canadian applications discussed, suggest a positive and interesting future for expert systems in planning whether it be only in the more technical sub-tasks or beyond. The usefulness of the tool and the general guidelines for assessing task suitability have been reinforced. Each application benefited from expert system features which conventional computer and manual decision-making methods could not offer. It seems justified at this stage to suggest that planning and expert systems do seem compatible in many situations and that expert systems are likely to become a widespread and extremely useful tool in the planning profession. Greater awareness of this new tool will help prepare planners to employ it wisely, avoiding its pitfalls and enjoying its benefits. REFERENCES Adler, Sy. (1987). The new information technology and the structure of planning practice. Journal of Planning and Education Research, 6, 2:93-98. Anderson, D. (1989). Artificial intelligence and intelligent, systems: The implications. Toronto: Halstead Press. Applied Artificial Intelligence Systems, Inc. & Cognos Advanced Technology. (1986). Expert systems: Their application in the Canadian transportation sector. Prepared for: Director General. Research and Development Transport Canada. Bluett, E. (1990). CMHC 1989 External Research Program Project: An expert system for RS-l/M permits in the city of Vancouver. Progress Report #2. Vancouver. Brail, Richard K. (1987). Microcomputers in urban planning and management New Brunswick. New Jersey: Center for Urban Policy Research. Brotchie, J., Hall, P., & Newton, P. (eds.) (1987). The spatial impact of technological cJiange. New York: Croom Helm. Brotchie, J., Newton, P., Hall, P., & Nijkamp, P. (eds.) (1985). The future of urban  form: The impact of new technology. New York: Nichols Publishing Co. Brown, D. F., & Schoen, D. A (1987). Using microcomputer C A D packages in planning. Journal of the American Planning Association. Spring. 249-258. Castello, M.(ed.) (1985). High technology, space and society society. Urban Affairs Annual Reviews, vol.28. London: Sage Publications. Catanese, A. J., & Snyder, J.C. (1988). Urban planning. 2nd edit Toronto: McGraw Hill Book Co. Chang, E., Clark, D., & Sidebottom, G . (undated). HERMES: Heuristic Emergency Response Management Expert System. Alberta Research Council. (Available from ARC) 131 / 132 Clark, D., Chang, E., & Sidebottom, G. (1988). HERMES: A prototype expert system for emergency response management. Emergency Preparedness Digest 15, 2:18-21. Charniak, E., McDermott, D. (1985). Introduction to artificial intelligence. Don Mills: Addison-Wesley Publishing Co. Cullen, I. (1986). Expert systems in planning analysis. Town Planning Review. 57, 3:239-251. Danziger, J. N., Dutton, W. H. , Kling, R., & Kraemer, K . L (1982). Computers & politics: High technology in American local governments. New York: Columbia University Press. Davis, J.R., & Grant I.W. (1987). ADAPT: A knowledge-based decision support system for producing zoning schemes. Environment and Planning Bj Planning and Design.14. 53-66. Davis, J.R., Campagnoni, P.T., & Nanninga, P.M. (1987). Roles for knowledge-based systems in environmental planning. Environment and Planning B: Planning and Design. 14, 239-254. Diamond, J.T., & Wright J-R. (1988). Design of an integrated spatial information system for multiobjective land-use planning. Environment and Planning B: Planning and Design. 15, 205-214. ESSA. Environmental and Social Systems Analysts Ltd. (1989?). SCREENER: A tool for decision making. (Promotional Material) Vancouver. File, P. (1988). Advice on available grants from an expert's assistant. The Planner. 74, 10:29-30. French, S. P. & Wiggins, L.L. Computer adoption and use in California planning agencies: Implications for education. Journal of Planning Education and Research. 8, 2:97-107. Garnham, A. (1988). Artificial intelligence: An introduction. New York: Routledge & Kegan Paul. / 133 Gar-On Yeh, A. (1988). Microcomputers in planning. Environment and Planning B: Planning and Design. 15, 237-239. Gar-On Yeh, A. (1988). Microcomputers in urban planning: applications, constraints, and impacts. Environment and Planning B: Planning and Design. 15, 241-254. Goshing, G . D. (1987). Application of expert systems in air traffic control. Journal of Transportation Engineering. 113 ,2:139-154. Han, S.Y., & Kim., T.J. (1989). Can expert systems help with planning? American Planning Association Journal summer, 296-308. Harris, R.A., Cohn, L.F., & Bowlby, W. (1987). Designing noise barriers using expert system CHINA. Journal of Transportation Engineering. 113, 2:127-138. Haugeland, J. (1985). Artificial intelligence: The very idea. Cambridge, Massachusetts: MIT Press. Hayes-Roth, F., Waterman, D., & Lenat, D. (1983). Building expert systems. Don Mills: Addison-Wesley Publishing Co., Inc. Henson, T. (ed.) of IBM Corp. (1988). Artificial intelligence and simulations: The diversity of applications. San Diego. California: SCS - The Society for Computer Simulation International. IEEE. Institute of Electrical and Electronics Engineering, Inc. (1989). 1989 Annual Artificial Intelligence Systems in Government Conference, ashington D.C: IEEE Computer Society Press. Kennedy, M . (1987). Logic, planning and computers: Prolog. Journal of Planning Education and Research. Association of Collegiate Schools of Planning, 7, 1:35-45. Kenyon, M.A.W. (1988). The Value of Spreadsheet Programs to Planners. M.A. Thesis. School of Community and Regional Planning. Vancouver: University of British Columbia. Kim, T.J., Wiggins, L.L., & Wright, J.R. (eds.) (1990).Expert systems applications to urban planning. Springer-Verlag. / 134 Klosterman, R.E. & Landis, J.D. {\%%).Microcomputers in U.S. planning: Past, present and future. Environment and Planning B: Planning & Design. 15, 355-367. Langendorf, R. (1985). Computers and decision making. Journal of the American Planning Association. Autumn, 422-433. Leary, M.E. (1986). Expert systems: What potential for planning? The Planner. 72, 12:38-39. Leary, M.E. (1988). Knowledge and reasoning in development control and urban design: An expert systems approach. Environment and Planning B- Planning and Design. 15, 383-398. Leary, M. , & Rodriguez-Bachiller, A. Expert systems in development control: Progress so far & the way ahead." The Planner. 174, 10:26-28. Lee, D. B. Jr. (1973). Requiem for large-scale models. American Institute of Planners Journal. May. Lima, R. J., (1985). Planning software survey. Planning Advisory Service Report no 388. Chicago: American Planning Association. Lopez-Mancisador del Rio, R. (1988). Madrid wins back its passengers. Developing metres 88. 36-37. Mendosa, M. , Masrani, R., Sutherland L., Sidebottom, G. , Chang, E., & Clark D. (1988). Proceedings from the Canadian conference on electrical and computer engineering. November 3-4. Vancouver: Alberta Public Safety Service and Alberta Research Council. Mick, S. & Wallace, W.A. (1986). Expert systems as decision aids for disaster management. Emergency Planning Digest 13, 3:16-20. Miller, R. K., & Walker, T.C. (1988). Artificial intelligence applications in engineering. Lilburn: The Fairmont Press, Inc. / 135 Moralee, D.S. (ed.) (1987). Research and Development in Expert System IV. Proceedings of Expert systems '87. The 7th Annual Technical Conference. December 14-17. Brighton: The British Computer Society Specialist Group on Expert Systems. Mulvey, D., & Skingle, B. (1988). Expert system aids fault diagnosis. Railway Gazette International. 144, 10:685. Nash, Kerry. (1986). The application of computers to planning tasks in the city of Sydney. Australian Planner. Journal of the Roval Australian Planning Institute. 24, 3:19-23. Newton, P.W. (1986). Microcomputers and the technology of planning. Australian Planner. Journal of the Roval Australian Planning Institute. 24, 3:5-11. Newton, P.W., & Taylor, M.A.P. (eds.) (1986). Microcomputers for local government and  management North Melbourne: Hargreen Publishing Co. Newton, P.W., Taylor, M.A.P., & Sharpe. R. (1988) Desktop Planning- Microcomputer applications for infrastructure and services planning. Melbourne: Hargreen Publishing Co. Newton, P.W. (1988). Microcomputer applications for urban infrastructure planning. Environment and Planning B: Planning and Design. 15, 255-268. Ortolano, L. & Perman, C D . (1987). A planner's introduction to expert systems. Journal of the American Planning Association. 53, 1:98-103. Ortolano, L. & Perman, C. (1990). Applications to urban planning: An overview, in Kim Wiggins and Wright 1990. Pfaffenberger, B. (1988). Microcomputer applications in qualitative research. Qualitative  Research Methods, vol. 14. Beverly Hills: Sage Publications. Research and Development Directorate. (1985). Proceedings of the Workshop on the Application of Expert Systems to Transportation. Montreal: Transport Canada. / 136 Robinson, V.B., Frank, A.U., & Blaze, M.A. (1986). Expert systems & geographic information systems: Review and prospects. Journal of Surveying Engineering. 112, 2:119-130. Robinson, V. B., Frank, A. U., & Blaze, M . A. (1986). 2>Expert systems applied to geographic information systems: Introduction, review and prospects. Computers-Environment and Urban Systems. 11, 4:161-173. Rodriguez-Bachiller, A. (1989). An expert system for the evaluation of local plans.?zpei presented at 13th Urban Data Management Symposium Lisbon (Portugal). May 29th - June 2nd. Rodriguez-Bachiller, A. (1990). Desktop planning: Can expert systems help? Paper presented at Third URSA-Net Advanced Seminar/Forum: Computers in Planning Patras. June. Sawicki, D. S. (1985). Microcomputer applications in planning. Journal of the American Planning Association. Spring, 209-215. Schon, D. A. (1983). The reflective practitioner: How professionals think in action. New York: Basic Books. Sivitanidou, R.M., & Polenske, K.R., (1988). Assesing regional economic impacts with microcomputers. Journal of the American Planning Association. Winter, 101-106. Tello, E (1988). Mastering AI tools and techniques. Indianapolis: Howard W. Sams and Company. Trivedi, M . M . (1989). Applications of artificial intelligence VII. Proceedings SPIE -International Society for Optical Engineering (SPIE). vol.1095. Pt.2. Orlando. Florida. Bellingham: SPIE Waterman, D. A. (1985). A guide to expert systems. Don Mills: Addison-Wesley Publishing Co. Wigan, M . R. (1987). Legal and ethical issues in expert systems used in planning. Environment and Planning B: Planning and Design 14, 306-321. / 137 Wiggins, L.L. (1986). Three low-cost mapping packages for microcomputers. Journal of the American Planning Association, Autumn, 480-488. Wohlwerth, N. (1987). Putting computer technology to work in emergency planning. Emergency Preparedness Digest.April-Tune 6-9. Yapa, L.S. (1988). Computer-Aided regional planning: A study in rural Sri Lanka. Environment and Planning R: Planning and Design. 15, 285-304. Young, W. (1988). Microcomputer-aided transport planning. Environment and Planning B: Planning and Design . 15, 269-283. APPENDIX A - S A M P L E LETTERS AND LISTS OF PROFESSIONALS C O N T A C T E D Sample Letter Sent to Canadian Planning SehooLs and Planners Listed on PLANET Professor Frank Palermo Head Department of Urban and Rural Planning Technical University of Nova Scotia P.O. Box 1000 Halifax, Nova Scotia, B3J 2X4 Lisa J. Colby May 18, 1990 Dear Professor Palermo, Please take a look at the attached letter (2 pages) and pass it on to the most appropriate faculty member. I am requesting information on the state of computer-based Expert Systems development at several research institutions, including all Canadian Planning Schools. I am also looking for suitable casestudies of Expert Systems (planning related) to explore in more detail. Even if no related work is ongoing within your faculty, please let me know that this is the case. I am most grateful for any cooperation that you can give me on this matter. Sincerely, Lisa Colby L J . Colby May 27, 1990 Dear X, In the process of investigating the state of computer-based Expert Systems applications work as it relates to planning in Canada, I am contacting universities and government research departments around the world which may be involved in such activities. I am a graduate student at the university of British Columbia (School of Community & Regional Planning) and am conducting thesis research on expert systems applications in 138 / 139 the planning field. • Has your department developed any prototype expert systems for planning-related problems? If so, how might I find out more about them? • Has your planning department conducted any research in the field? • Has your department produced any papers relating to such work and, if so, how might I obtain copies? • Does your curriculum include anything on expert systems? If so, to what extent (briefly)? • Finally, if unable to help with any of the above questions, do you know of anyone who is doing research in this area? Any information you could send me in this regard would be most useful and appreciated. If you would prefer to respond through electronic mail, my e-mail addresses for Internet and BITNET are: Internet COLBYMtsg.ubc.ca BITNET userSY05Ubcmtsg Response via this channel would be useful to me both in terms of time savings and convenience. Alternatively, my postal addresses and research advisor are: Lisa J. Colby c/o School of Community and Regional Planning (address as per letterhead) Research Advisor: Prof. H.C. Hightower ...(604) 228-5977 messages for L. Colby or Prof. Hightower ...(604) 228-3276 phone Any suggestions regarding my research would be welcome. Thankyou in advance for any attention you can give to this letter. Sincerely yours, Lisa J. Colby / 140 Sample letter Sent to Authors of Expert Systems Portia File Department of Maths and Computer Studies Dundee College of Technology U.K. Lisa J. Colby May 24, 1990 Dear Ms. File, Your work on G R A N T ADVISOR came to my attention in an article published in The Planner (October, 1988). Presently a graduate student at the University of British Columbia (School of Community and Regional Planning), I am researching expert systems application in the planning field. Your project may be a useful case-study for my thesis work. I am looking for examples of expert systems to illustrate the potential of expert system use in planning. I am also curious to know of people's experiences in the development and use of these systems. This would assist me in developing guidelines to assess whether a given planning task is suitable for expert system use or not I would appreciate your help with some questions regarding your work. I am interested in your views, opinions and experiences of the G R A N T ADVISOR project If you are able to forward me any documentation or publications, many of these questions may be answered within their text and I would welcome that form of response. I understand the value of your time and encourage you to respond separately only to those questions which you feel are relevant to your experience and which are otherwise unaddressed in the documentation. 1. W H E R E MIGHT I FIND D O C U M E N T A T I O N O F GRANT ADVISOR ? 2. C A N I G E T F R O M Y O U : DISKS, REPORTS, OR EXAMPLES OF GRANT ADVISOR'S APPLICATIONS TO R E A L PROBLEMS? 3. a) W H A T H A R D W A R E DID Y O U USE T O DEVELOP T H E GRANT ADVISOR PROGRAM? b) What hardware does GRANT ADVISOR run on? a) W H A T SOFTWARE DID Y O U USE TO DEVELOP T H E GRANT ADVISOR PROGRAM? b) What software does G R A N T ADVISOR run on? / 141 5. a) W H A T TYPE O F PROBLEMS C A N G R A N T ADVISOR SOLVE? b) Is the final version of G R A N T ADVISOR capable of performing all of the work for which the project was originally conceived? c) If not, what was compromised and why? 6. a) W H A T ATTRIBUTES OF T H E PROBLEM M A K E IT M O R E SUITABLE TO EXPERT SYSTEM APPLICATION T H A N OTHER METHODS C U R R E N T L Y BEING USED? b) What are the weaknesses in the current system of handling the job (before G R A N T ADVISOR)? c) Are human personnel capable of doing as good a job as the expert system, in this particular case? 7. a) HAS T H E PERFORMANCE O F G R A N T ADVISOR SATISFIED Y O U R EXPECTATIONS? b) If NOT, would you attribute the problems to (check those to which the answer is yes and elaborate if necessary): The negative attitude of users? Difficulty on the part of human experts in declaring their decision making process? Difficulty on the part of programmers in adequately representing the experts' declared decision making process? An insufficiently complete knowledge-base. Inherent unsuitability of the task to expert system technology (please elaborate). Other (please elaborate) c) Were there any unexpected benefits or drawbacks to the use of G R A N T ADVISOR for the intended tasks? If so, please explain. If you would prefer to respond through electronic mail, my e-mail addresses on Internet and BITNET are: Internet: ColbyMtsg.ubc.ca BITNET: usersy05Ubc.mts.g / 142 Response via this channel would be useful both in terms of time savings and convenience. If you wish to make any suggestions regarding my research, please feel free. Thankyou in advance for any attention you can give to this letter. Sincerely yours, L.J. Colby / 143 Sample letter Sent to Software Companies IntelliCorp 1975 El Camino Real West Mountain View C A 94040 Lisa Colby May 18, 1990 Dear Sir/ Madam, IntelliCorp has come to my attention as a company which may be developing or handling commercially available Expert Systems. I am a graduate student at the university of British Columbia (School of Community and Regional Planning) and am researching Expert Systems applications in the planning field. * Do you have a catalogue or list of commercially available exert systems and expert system shells which might relate to the urban planning field? * Could you suggest government agencies or planning firms which are currently using such technology and whom I might approach for feedback on their experiences? * Can you suggest an expert system which might serve as a useful case-study for more detailed coverage in my research, and if so where might I find information on it? Any information you could send me about available technology in the planning domain, would be useful. Thankyou in advance for any attention you can give to this letter. Sincerely yours, L.J. Colby List of Canadian Planning Schools Contacted Professor G. Bargh, Head Department of City Planning University of Manitoba Winnipeg, Manitoba R3T 2N2 Professor W. Jamieson, Director Faculty of Environmental Design University of Calgary Calgary, Alberta T2N 1N4 Professor D. Douglas, Director School of Rural Planning And Development University of Guelph Guelph, Ontario N I G 2W1 Professor M . Gertler Programme in Planning Department of Geography University of Toronto Toronto, Ontario M5S 1A1 Professor G. Daly Faculty of Environmental Studies Room 349, Lumbers Building York University 4700 Keele St North York, Ontario M3J 1P3 Professor M . Qadeer, Director School of Urban and Regional Planning Macintosh-Corry Hall Queen's University Kingston, Ontario K7L 3N6 Professor R. Keeble, Director School of Urban and Regional Planning Ryerson Polytechnical Institute 350 Victoria SL Toronto, Ontario M5B 2K3 M . E Weiss-Altaner, Directeur Departement d'etudes urbaines Universite du Quebec a Montreal C P . 8888 Montreal, Quebec H3C 3P8 M . P. Frechette, Directeur Programme, ATDR Pavilion Felix-Antoine-Savard Vureau 1624 Universite Laval Ste. Foy, Quebec GIK 7P4 Prof. B. Smith, Head Environmental Planning Department Nova Scotia College of Art and Design 5163 Duke St Halifax, Nova Scotia B3J 3J6 Prof. J.C. Stabler, Coordinator Urban and Regional Planning Program University of Saskatchewan Saskatoon, Sask. S7N 0W0 M . R. Parenteau, Directeur Institut d'Urbanism Universite de Montreal CP 6128, Succ "A" Montreal, Quebec H3C 3J7 Professor F. Palermo, Head Department of Urban and Rural Planning Technical University of Nova Scotia P.O. Box 1000 Halifax, N.S. B3J 2X4 Professor J. Wolfe, Directeur School of Urban Planning McGill University 815 Sherbrooke St W. Montreal, Quebec H3A 2K6 Professor L. Martin, Director School of Urban and Regional Planning University of Waterloo Waterloo, Ontario N2L 3G1 List of Planners Contacted Through PLANET Electronic Mail Network David Barkin, Univ. Autonoma Metropolitana (Morelia, Mexico) Mike Batty, Univ. of Wales Earl Bell, Univ. of Washington John Bis, SUNY Buffalo Dick Brail, Rutgers Univ. David Brown, McGill Univ. Naomi Carmon, Israel Institute of Technology Tim Cartwright, York Univ. Cheryl Contant, Univ. of Iowa John Cook, Univ. of North Carolina Peter Doan, Univ. of North Carolina Bill Drummond, Georgia Institute of Technology Ken Dueker, Portland State Univ. Yair Etzion, Ben Gurion Univ. of the Negev Allan Feldt, Univ. of Michigan Joe Ferreira, MIT Jim Fisher, Univ. of Cincinatti John Fitzsimons, Univ. of Guelph Jon D. Fricker, Purdue, Univ. Alex Gabbour, Univ. de Montreal Brent Hall, Univ. of Waterloo Eric Heikkila, Univ. of Southern California Lewis Hopkins, Univ. of Illinois Dave Johnson, Univ. of Tennessee T. John Kim, Univ., of Illinois Dick Klosterman, Univ. of Akron Mickey Lauria, Univ. of New Orleans Jim Mars, Ryerson Polytechnical Institute Ross Newkirk, Univ. of Waterloo Bill Page, Univ. of Wisconsin-Milwaukee David Philips, Univ. of Virginia Duane Shinn, Iowa State Univ. Lyna Wiggins, MIT / 148 List of Expert Systems Authors to Whom letters/ Inquiries Sent Marc P. Armstrong Dept. of Geography & Computer Science Univ. of Iowa, Iowa City, IA (A Rule-based Advisory System for Ground Water Quality Assessment at Hazardous Waste Disposal Sites) Ernest Chang (HERMES) Head, Advanced Computing and Engineering Alberta Research Council 6815 8th St, Calgary T2E 7H7 Don Clark (HERMES) Senior Compliance Officer Client Information Center Alberta Public Safety Services 1032 146 St Edmonton, T5N 3A2 J.R. Davis (ADAPT) Division of Water & Land Resources CSIRO, GPO Box 1666, Canberra A C T 2601 Australia Portia File (GRANT ADVISOR) Department of Maths and Computer Studies Dundee College of Technology U.K. I.W. Grant (ADAPT) Division of Water & Land Resources CSIRO, GPO Box 1666, Canberra ACT 2601 Australia Michael Kennedy College of Architecture Pense Hall Univ. of Kentucky Lexington, K Y 40506 / 149 M.E. Leary (Knowledge & Reasoning in Development Control and Urban Design) Dept. of Town Planning Oxford Polytechnic Headington, Oxford 0X3 OBP, England Arif Merchant c/o School of Planning University of Cincinnati Cincinnati, O H 45521-0016 Peter Newton (Several systems) CSIRO P.O. Box 56 Highett 3190, Melbourne, Australia Greg Sidebottom (HERMES) Programmer in Artificial Intelligence Advanced Computing and Engineering Alberta Research Council 6815 8th St, Calgary T2E 7H7 Gar-On Yeh Centre of Urban Studies & Urban Planning University of Hong Kong Pokfulam Rd., Hong Kong List of AI Service and Software Companies to Whom Letters Sent Peter F. Rowat Coast Mountain Intelligence 1826 West 1st Ave. Vancouver, B.C., V6J 1G5 Lynn Lord Marketing Coordinator MacDonald Dettwiler and Associates, Ltd. 3751 Shell Rd. Richmond, B.C., V6X 2Z9 Teknowledge, Inc. 525 University Ave. Palo Alto, California 94301 UMECORP 275 Magnolia Ave. Larkspur, California, 94939 Production Systems Technologies 642 Gettysburg St, Pittsburgh PA 15206 Expert Systems International 1150 First Avenue King of Prussia, PA 19406 Inference Corporation 5300 West Century Blvd., 5th floor Los Angeles, C A 90045 IntelliCorp 1975 El Camino Real West Mountain View C A 94040 Level 5 Research Inc. 4980 South A - A - A Melbourne Beach, F L 32951 Walter Berndl Logicware, Int 204-2065 Dundas E Mississauga, Ontario, Canada L4X 2W1 • i i i i i i i i i i i i i i i i i i i i i i m i n . 6 g^MSBg 4 B I R e s p o n s e M a n a 9 e m e n l C , l 1 1 3 ° 3 »»«••>.» l>... ^ J , ^ . tj.V,...; E x p e r t S y s t e m E G tut Bfl9 _ 0 u *\ WATCH THESE LEVELS E v a c a a t l t m D l i l u c a S l l H A i U t t D u p r E x p U i l * * D i n g e r tirm D**f«r S l t t D*»£tr P k i k l l c D**g«r W«rker D u i f c r E f t v l r a a n a a t D u g i r •C •C 110 0 r HA I c SUGGESTED THINGS TO OO: Fry Identifying the a c t i v i t y type. Iry identifying the d i s t r i c t type. ADVISED EMERGENCY ACTION GIVEN: Renove a l l Injured people fron truck atte. Prevent further access to truck aIte. To hake the envlronnent aafe, a l l Ignition sources should be e l l n l n AOVISEO EMERGENCY ACTION TAKEN: Diotrlct Typo: OnTinown L2: w i n d o w r n d r i u . llhu 3 flo9 i:30:l/J LISPrt Sleep APPENDIX C SAMPLE REPORTS FROM SCREENER SCREENER - Detailed Report Transport Canada Transports Canada Ai r p o r t s A u t h o r i t y Group Groupe de g e s t i o n des aeroports PROJECT REGISTER AND DETAILED SCREENING REPORT Date Examined: 18/03/90 ESS Version: Ver 2.0d Region: Western Location: Calgary I n t ' l Screening Report No.: Project No.: S2P203-0034 TEC: 1542.4 F i s c a l Year: 1993/94 Project T i t l e : Demonstration Screening Project Project D e s c r i p t i o n : Northwards Extension of Main Runway DETAILED RATIONALE FOR THE IMPACT CONCLUSION ON EACH COMPONENT T e r r e s t r i a l Animals (NO IMPACT) T e r r e s t r i a l Habitat (NO IMPACT) Vegetation (MITIGATED OVERALL): Grasses (MITIGATED In General): The p o t e n t i a l impact: equipment use i n v o l v i n g frequent passage and off-pavement t r a v e l w i l l create dust and adversely a f f e c t A i r p o r t Grassed Areas (grasses) can be mi t i g a t e d (see l i s t of m i t i g a t i o n actions below). The p o t e n t i a l impact: hau l i n g i n v o l v i n g off-pavement t r a v e l and frequent passage w i l l cover vege t a t i o n with dust and damage A i r p o r t Grassed Areas (grasses) can be m i t i g a t e d (see l i s t of m i t i g a t i o n actions below). A p o t e n t i a l impact of t h i s p r o j e c t i s : b l a s t i n g / d r i l l i n g w i l l adversely a f f e c t A i r p o r t Grassed Areas (grasses). This impact w i l l l i k e l y be IN-SIGNIFICANT because: - A i r p o r t Grassed Areas (grasses) i s / a r e not s p e c i a l status. A p o t e n t i a l impact of t h i s p r o j e c t i s : h a u l i n g i n v o l v i n g 152 DETAILED RATIONALE cont'd frequent passage and off-pavement travel w i l l destroy or damage Airport Grassed Areas (grasses). This impact w i l l l i k e l y be IN-SIGNIFICANT because: - Airport Grassed Areas (grasses) is/are not special status. A potential impact of this project i s : vegetation-removal w i l l destroy or degrade Airport Grassed Areas (grasses) . This impact w i l l l i k e l y be IN-SIGNIFICANT because: - Airport Grassed Areas (grasses) is/are not special status. Crops (MITIGATED In General) : The potential impact: equipment use involving frequent passage and off-pavement travel w i l l create dust and adversely affect Airport Crop Leases (crops) can be mitigated (see l i s t of mitigation actions below). The potential impact: hauling involving off-pavement travel and frequent passage w i l l cover vegetation with dust and damage Airport Crop Leases (crops) can be mitigated (see l i s t of mitigation actions below). A potential impact of this project i s : b l a s t i n g / d r i l l i n g w i l l adversely affect Airport Crop Leases (crops). This impact w i l l l i k e l y be IN-SIGNIFICANT because: - Airport Crop Leases (crops) is/are not special status. A potential impact of this project i s : hauling involving frequent passage and off-oavement travel w i l l destroy or damage Airport Crop Leases (crops). This impact w i l l l i k e l y be IN-SIGNIFICANT because: - Airport Crop Leases (crops) is/are not special status. A potential impact of this project i s : vegetation removal w i l l destroy or degrade Airport Crop Leases (crops). This impact w i l l l i k e l y be IN-SIGNIFICANT because: - Airport Crop Leases (crops) is/are not special status. Aquatic Animals (IN-SIGNIFICANT OVERALL): Aquatic Birds (IN-SIGNIFICANT In General): A potential impact of this project i s : b l a s t i n g / d r i l l i n g w i l l adversely affect Shorebirds (aquatic birds). This imoact w i l l l i k e l y be IN-SIGNIFICANT because: . - Shorebirds (aquatic birds) is/are not special status. DETAILED RATIONALE conc'd / A potential impact of this project is: vegetation removal involving large area w i l l reduce riparian habitat and decrease the abundance of Shorebirds (aquatic birds). This impact w i l l l i k e l y be IN-SIGNIFICANT because: - Shorebirds (aquatic birds) is/are not special status. A potential impact of this project i s : a i r c r a f t movement involving frequent passage and nef > 30 w i l l create noise and adversely affect Waterfowl (aquatic birds) . This impact w i l l l i k e l y be IN-SIGNIFICANT because: - Waterfowl (aquatic birds) is/are not special status. A potential impact of this project i s : b l a s t i n g / d r i l l i n g w i l l adversely affect Waterfowl (aquatic birds). This impact w i l l l i k e l y be IN-SIGNIFICANT because: - Waterfowl (aquatic birds) is/are not special status. A potential impact of this project i s : equipment use involving high noise l e v e l w i l l create noise and adversely affect Waterfowl (aquatic birds). This impact w i l l l i k e l y be IN-SIGNIFICANT because: - equipment use is/are not high noise l e v e l . - Waterfowl (aquatic birds) is/are not special status. A potential impact of this project i s : vegetation removal involving large area w i l l reduce riparian habitat and decrease the abundance of Waterfowl (aquatic birds). This impact w i l l l i k e l y be IN-SIGNIFICANT because: - Waterfowl (aquatic birds) is/are not special status. Aquatic Habitat (NO IMPACT) Surface Waters (OVERRIDDEN: IN-SIGNIFICANT) : The rationale for this override i s : The nearby surface waters to this project are very seasonal, and unlikely to be s i g n i f i c a n t l y adversely affected. Any potential impacts previously calculated by the system (and l i s t e d below, i f any) for surface waters w i l l be superceded by th i s override. Reservoirs (SIGNIFICANT In General) : A potential impact of this project i s : grading ( c u t / f i l l ) involving large area w i l l create sedimentation and decrease the quality of North Retention Pond (reservoirs). This impact w i l l l i k e l y be SIGNIFICANT because: - grading ( c u t / f i l l ) does involve large area. - grading ( c u t / f i l l ) w i l l occur where drainage i s into the North Retention Pond. The potential impact: asphalting/concreting involving DETAILED RATIONALE cont'd large area w i l l alter the flow regime of North Retention Pond (reservoirs) can be mitigated (see l i s t of mitigation actions below. The potential impact: asphalting/concreting involving large area w i l l increase contaminant runoff and reduce the quality of North Retention Pond (reservoirs) can be mitigated (see l i s t of mitigation actions below). The potential impact: hauling involving frequent passage and off-pavement travel w i l l create dust and decrease the quality of North Retention Pond (reservoirs) can be mitigated (see l i s t of mitigation actions below). The potential impact: topsoil stripping involving large area w i l l increase sedimentation and decrease the quality of North Retention Pond (reservoirs) can be mitigated (see l i s t of mitigation actions below). The potential impact: vegetation removal involving large area w i l l increase run-off and increase the peak flow of North Retention Pond (reservoirs) can be mitigated (see l i s t of mitigation actions below). Groundwaters (NO IMPACT) Landforms/Terrain (NO IMPACT) Soils (MITIGATED OVERALL): The potential impact: equipment use involving heavy equipment and off-pavement travel and frequent passage w i l l compact Airport S i l t y Clays (soils) can be mitigated (see l i s t of mitigation actions below). The potential impact: topsoil stripping involving large area w i l l decrease the productivity of Airport S i l t y Clays (soils) can be mitigated (see l i s t of mitigation actions below) . The potential impact: vegetation removal involving large area w i l l increase wind erosion and lead to the loss of Airport S i l t y Clays (soils) can be mitigated (see l i s t of mitigation actions below. The potential impact: vegetation removal involving large area w i l l lead to the movement of Airport S i l t y Clays (soils) can be mitigated (see l i s t of mitigation actions below). A potential impact of this project i s : grading ( c u t / f i l l ) involving road cuts w i l l lead to erosion and the movement of Airport S i l t y Clays ( s o i l s ) . This imoact w i l l l i k e l y be IN-SIGNIFICANT because: - grading ( c u t / f i l l ) is/are not road cuts. Atmosphere (IN-SIGNIFICANT OVERALL): DETAILED RATIONALE cont'd / 1 5 6 Airshed (IN-SIGNIFICANT In General): A potential impact of this project i s : vegetation removal involving burning and large area w i l l create smoke and reduce v i s i b i l i t y of Airport Airshed (airshed). This impact w i l l l i k e l y be IN-SIGNIFICANT because: - vegetation removal is/are not burning. - Airport Airshed (airshed) is/are not stable conditions. Resource Harvests (NO IMPACT) Recreation (NO IMPACT) Community Facilities/Services (NO IMPACT) Land Use (SIGNIFICANT OVERALL): Agri c u l t u r a l Land Use (SIGNIFICANT In General): A potential impact of this project i s : a i r c r a f t movement involving nef > 30 and frequent passage w i l l create noise and decrease options for N . Cattle Farms (agricultural land use) . This impact w i l l l i k e l y be SIGNIFICANT because: - a i r c r a f t movement does involve nef > 30 and frequent passage. - N . Cattle Farms is/are o f f s i t e . - a i r c r a f t movement w i l l occur overhead of the N. Cattle Farms. Transportation/Util Corridors (SIGNIFICANT In General): A potential impact of this project i s : a i r c r a f t movement involving nef > 30 and frequent passage w i l l create noise and decrease options for Deerfoot T r a i l N.E. (transportation/utii corridors). This impact w i l l l i k e l y be SIGNIFICANT because: - a i r c r a f t movement does involve nef > 30 and frequent passage. - Deerfoot T r a i l N.E. is/are o f f s i t e . - a i r c r a f t movement w i l l occur overhead of the Deerfoot T r a i l N.E. Employment/Economy (NO IMPACT) People (NO IMPACT) MITIGATION ACTIONS SELECTED / 157 Below are l i s t e d the 8 m i t i g a t i o n actions selected as part of t h i s screening i n c l u d i n g a more d e t a i l e d d e s c r i p t i o n for each mitigation, and one or more a p p l i c a b l e g u i d e l i n e s . Selected M i t i g a t i o n s : Avoiding s t o c k p i l i n g of s o i l f o r long periods. D e s c r i p t i o n : Avoid s t o c k p i l i n g t o p s o i l f o r lengthy periods of time. Applicable G u i d e l i n e s : Transport Canada, A i r p o r t F a c i l i t i e s Branch, "Environmental Impact Studies", AK-75-02-003, March (1981). Selected M i t i g a t i o n s : Containing and d i r e c t i n g surface discharge to waste water treatment system. D e s c r i p t i o n : A i r f i e l d drainage systems should have the c a p a b i l i t y , when required, to channel a i r p o r t run-off to one l o c a t i o n f o r waste treatment processing. Landscaped f i l t e r berras around areas at points of discharge could be constructed d i r e c t i n g surface discharge to a municipal treatment f a c i l i t y or storage lagoon. Applicable G u i d e l i n e s : Transport Canada, A i r p o r t F a c i l i t i e s Branch, "Environmental Impact Studies", AK-75-02-003, March (1981). Environment Canada, Environmental Protection Service, "Guidelines f o r E f f l u e n t Q u a l i t y and Waste Mater Treatment, EPS l-SC-76-1, 17pp., A p r i l (1976). Environment Canada, Environmental Protection Service, "Interim Guidelines f o r Waste Water Disposal i n Northern Canada Communities, EPS 2-WP-74-1, 10pp., JuneU984). F i s h e r i e s and Oceans, "Urban Development: Guidelines f o r P r o t e c t i o n of F i s h Habitat i n Insular Newfoundland", 95pp., March (1983) . Selected M i t i g a t i o n s : L i m i t i n g the maximum area uncovered and exposed at any one time. D e s c r i p t i o n : Because of the s i g n i f i c a n t amount of ground cover removal a s s o c i a t e d with a i r p o r t s (runways, buildings, and roads, e t c . ) , the maximum area uncovered and exposed at any one time should be l i m i t e d . At present there does not appear to be any s p e c i f i e d maximum allowable area subject to exposure at one time. The rate of erosion, however, varies with slope, s o i l type, p r e c i p i t a t i o n , wind (speed, d i r e c t i o n , and duration), and the season. Consequently, a f l e x i b l e plan f o r erosion and sediment c o n t r o l should be i n i t i a t e d p r i o r to and during c o n s t r u c t i o n a c t i v i t i e s . A pplicable G u i d e l i n e s : Transport Canada, A i r p o r t F a c i l i t i e s Branch, Measures to M i t i g a t e and Ameliorate the Adverse Effects on the P h y s i c a l •i SELECTED MITIGATIONS cont'd / 158 Environment (Excluding A i r c r a f t Noise) By A i r p o r t Expansion, Development and Operations", 60 pp. February (1977) . Environment Canada, Environmental Protection Service, "Environmental Code of Good P r a c t i c e for General Construction, EPS l-EC-80-1, 51 pp., March (1980). Transport Canada, A i r p o r t F a c i l i t i e s Branch, "Environmental Impact Studies", AK-75-02-003, March (1981). Selected M i t i g a t i o n s : Not pushing or dumping f i l l i n t o water bodies. D e s c r i p t i o n : Do not push or dump any type of f i l l into streams. Applicable Guidelines: Transport Canada, A i r p o r t F a c i l i t i e s Branch, "Environmental Impact Studies", AK-75-02-003, March (1981). Selected M i t i g a t i o n s : Revegetating and sodding. D e s c r i p t i o n : As much landscaping and sodding as possible should occur during the construction phase, e s p e c i a l l y when the a c t i v i t y i nvolves the a l t e r a t i o n of ground cover. Applicable Guidelines: Transport Canada, A i r p o r t F a c i l i t i e s Branch, Measures to M i t i g a t e and Ameliorate the Adverse E f f e c t s on the P h y s i c a l Environment (Excluding A i r c r a f t Noise) By A i r p o r t Expansion, Development and Operations", 60 pp. February (1977) . Environment Canada, Environmental Protection Service, "Environmental Code of Good Pr a c t i c e for General Construction, EPS l-EC-80-1, 51 pp., March (1980). Transoort Canada, A i r o o r t F a c i l i t i e s Branch, "Environmental Impact Studies", AK-75-02-003, March (1981). Selected M i t i g a t i o n s : R e s t r i c t i n g speed and a c c e l e r a t i o n on haul roads. D e s c r i p t i o n : R e s t r i c t speed and excessive a c c e l e r a t i o n on haul roads. A p p l i c a b l e Guidelines: Transport Canada, A i r p o r t F a c i l i t i e s Branch, Measures to M i t i g a t e and Ameliorate the Adverse E f f e c t s on the P h y s i c a l Environment (Excluding A i r c r a f t Noise) By A i r p o r t Expansion, Development and Operations", 60 pp. February (1977) . Environment Canada, Environmental Protection Service, "Environmental Code of Good P r a c t i c e for General Construction, EPS l-EC-80-1, SI pp., March (1980). Transport Canada, A i r p o r t F a c i l i t i e s Branch, "Environmental SELECTED MITIGATIONS cont'd / 159 Impact Studies", AK-75-02-003, March (1981). Selected Mitigations: Directing stormwater to storage lagoons. Description: Stormwater should be directed to storage lagoons. Applicable Guidelines: Transport Canada, Airport F a c i l i t i e s Branch, Measures to Mitigate and Ameliorate the Adverse Effects on the Physical Environment (Excluding A i r c r a f t Noise) By Airport Expansion, Development and Operations"/ 60 pp. February (1977) . Transport Canada, Airport F a c i l i t i e s Branch, "Environmental Impact Studies", AK-75-02-003, March (1981). Fisheries and Oceans, "Urban Development: Guidelines for Protection of Fish Habitat in Insular Newfoundland", 95pp., March (1983) . Transport Canada, Airport F a c i l i t y Branch," Manual of Environmental Protection: Design and Construction -Southern Canada", AK-75-04-000, II pp., March (1983). Transport Canada, Airport F a c i l i t y Branch," Manual of Environmental Protection: Northern Canada", AK-75-03-000, 12 pp., December (1983). Transport Canada, Airport F a c i l i t i e s Branch, "Storm Water Pollution Control Manual", AK-65-03-000, March (1985) . Selected Mitigations: Treating haul roads to reduce dust. Description: Designated haul roads should be treated to reduce potential dusting of vegetation. Applicable Guidelines: Transport Canada, Airport F a c i l i t i e s Branch, Measures to Mitigate and Ameliorate the Adverse Effects on the Physical Environment (Excluding A i r c r a f t Noise) 3y Airport Expansion, Development and Operations", 60 pp. February (1977) . Environment Canada, Environmental Protection Service, "Environmental Code of Good Practice for General Construction, EPS l-EC-30-1, 51 pp., March (1980). Transport Canada, Airoort F a c i l i t i e s Branch, "Environmental Impact Studies", AK-75-02-003, March (1981). / 160 - Sample Cross-Impact Matrix Report Cross-Impact Matrix Report Page: 1 Date: 18/03/90 Time: 11:26 Transport Canada A i r p o r t s Authority Group Transports Canada Groupe de gestion des aeroports PROJECT REGISTER AND CROSS-IMPACT MATRIX Date Examined: 18/03/90 ESS Version: Ver 2.Od Region: Western Location: Calgary I n t ' l Screening Report No.: Project No.: S2P203-0034 TEC: 1542.4 F i s c a l Year: 1993/94 Project T i t l e : Demonstration Screening Project Project Description: Northwards Extension of Main Runway LIST OF SELECTED ACTIVITIES Code A c t i v i t y 1 - a i r c r a f t movement 2 - asphalting/concreting 3 - b l a s t i n g / d r i l l i n g 4 - culvert i n s t a l l a t i o n 5 - equipment use 6 - fencing 7 - grading ( c u t / f i l l ) 8 - hauling 9 - placing subgrade materials 10 - t o p s o i l s t r i p p i n g 11 - vegetation removal Cross Impact (cont'd) / 161 Component T e r r e s t r i a l mammals T e r r e s t r i a l mammals T e r r e s t r i a l mammals T e r r e s t r i a l birds T e r r e s t r i a l mammal habitat T e r r e s t r i a l mammal habitat T e r r e s t r i a l mammal habitat T e r r e s t r i a l b i r d habitat T e r r e s t r i a l b i r d habitat Trees Trees Shrubs Grasses Crops Aquatic b i r d s Aquatic birds Aquatic b i r d habitat Aquatic b i r d habitat Aquatic b i r d habitat Streams Streams Streams Reservoirs Reservoirs Underground aquifers S o i l s S o i l s S o i l s Airshed Airshed Airshed Airshed F i r e f i g h t i n g F i r e f i g h t i n g P o l i c e Drinking water Sanitary sewers Sanitary sewers Sanitary 3ewers Sanitary sewers Storm sewers Storm sewers Storm sewers Storm sewers Power l i n e s Gaslines Public transportation Public transportation Roadways Roadways Roadways Name Jackrabbits Badgers Coyotes Large Raptors Coyote Denning Sites A i r p o r t Grassed Areas Ai r p o r t Crop Leases A i r Terminal B u i l d i n g N. Retent. Low Areas Barlow Rd. Orn. Groves Old Tree Nursery A.T. Landscape Shrubs A i r p o r t Grassed Areas A i r p o r t Crop Leases Waterfowl Shorebirds North Retention Pond Taxiway F Ditch F i r e Tr. Area Ditch Tank Farm Ditch East Central Ditch Cargo Area Ditch North Retention Pond S. Central Surge Pond Air p o r t Aquifer Gulf O i l S i t e Contaminated S o i l S i t e A i r p o r t S i l t y Clays East Comm'ty Airshed West Comm'ty Airshed South Comm'ty Airshed A i r p o r t Airshed F i r e h a l l 27 (South) F i r e h a l l 13 (North) Ai r p o r t RCMP 80th Ave Acreage Wells A.C. Tower Septic Pond West Hangar Sewer Line East Hangar Sewer Line A.T./Cargo Forced Mn. S.W. Storm Sewer S.C. Storm Sewer S .E . Storm Sewer N. Retent. Storm Sewer Mn Serv Corridor Power Mn Serv Corridor Gas Taxicab Services Shuttle Services Deerfoot T r a i l N.E. McCail Way McKnight Blvd. A c t i v i t y Code 1 2 3 4 5 N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N M M I M / 162 Cross Impact (cont'd) Component Name A c t i v i t y Code 1 2 3 4 5 Roadways - Barlow T r a i l N.E. N A g r i c u l t u r a l land use - A i r p o r t Crop Leases N A g r i c u l t u r a l land use - N. C a t t l e Farms Commercial land use - East Light Commercial N Commercial land use - East Hangar Commercial N Commercial land use - West Light Commercial N Commercial land use - West Hangar Commercial N Commercial land use - Gen. Cargo Area N Commercial land use - A i r Term. Commercial N Recreational land use - McCall Lk. Golf Course N Res i d e n t i a l land use - C i t y of Calgary N T r a n s p o r t a t i o n / u t i l c o r r i d o r s - Deerfoot T r a i l N.E. Tra n s p o r t a t i o n / u t i l c o r r i d o r s - Main Serv Corridor N Employment - Contruction Employment N Employment - Maintenance Employment N Employment - Operations Employment N Employment - Service Employment N Regional economic a c t i v i t y - C i t y of Calgary N Adjacent landowners - N. C a t t l e Farms N Adjacent landowners - 80th Ave. Acreages N A i r l i n e passengers - Private Passengers N A i r l i n e passengers - A i r l i n e Passengers N Air p o r t workers - A l l A i r p o r t Personnel N Re s i d e n t i a l neighborhoods - W. Comm'ty Residents N Re s i d e n t i a l neighborhoods Residential neighborhoods -S. Comm'ty Residents E. Comm'ty Residents N - N General population - General Population N / 163 Cross Impact (cont'd) Component Name A c t i v i t y Code 6 7 8 9 10 T e r r e s t r i a l mammals - Jackrabbits N T e r r e s t r i a l mammals - Badgers N T e r r e s t r i a l mammals - Coyotes N T e r r e s t r i a l b i r d s - Large Raptors N T e r r e s t r i a l mammal habitat - Coyote Denning Sites N T e r r e s t r i a l mammal habitat - A i r p o r t Grassed Areas N T e r r e s t r i a l mammal habitat - A i r p o r t Crop Leases N T e r r e s t r i a l b i r d habitat - A i r Terminal Building N T e r r e s t r i a l b i r d habitat - N. Retent. Low Areas N Trees - Barlow Rd. Orn. Groves N Trees - Old Tree Nursery N Shrubs - A.T. Landscape Shrubs N Grasses - Airport Grassed Areas Crops - Airport Crop Leases Aquatic b i r d s - Waterfowl Aquatic b i r d s - Shorebirds Aquatic b i r d habitat - North Retention Pond N Aquatic b i r d habitat - Taxiway F Ditch N Aquatic b i r d habitat - F i r e Tr. Area Ditch N Streams - Tank Farm Ditch N Streams - East Central Ditch N Streams - Cargo Area Ditch N Reservoirs - North Retention Pond Reservoirs - S. Central Surge Pond N Underground aquifers - Airport Aquifer N S o i l s - Gulf O i l S i t e N S o i l s - Contaminated S o i l S i t e N S o i l s - Airport S i l t y Clays Airshed - East Comm'ty Airshed N Airshed - West Comm'ty Airshed N Airshed - South Comm'ty Airshed N Airshed - Airport Airshed F i r e f i g h t i n g - F i r e h a l l 27 (South) N F i r e f i g h t i n g - F i r e h a l l 13 (North) N P o l i c e - Airport RCMP N Drinking water - 80th Ave Acreage Wells N Sanitary sewers - A.C. Tower Septic Pond N Sanitary sewers - West Hangar Sewer Line N Sanitary sewers - East Hangar Sewer Line N Sanitary sewers - A.T./Cargo Forced Mn. N Storm sewers - S.W. Storm Sewer N Storm sewers - S.C. Storm Sewer N Storm sewers - S.E. Storm Sewer N Storm sewers - N. Retent. Storm Sewer N Power l i n e s - Mn Serv Corridor Power N Gaslines - Mn Serv Corridor Gas N Pubiic transportation - Taxicab Services N Public transportation - Shuttle Services N Roadways - Deerfoot T r a i l N.E. N Roadways - McCall Way N Roadways - McKnight Blvd. N / 164 Cross Impact (cont'd) Component Name A c t i v i t y Code 6 7 8 9 10 Roadways - Barlow T r a i l N.E. N A g r i c u l t u r a l land use - Airport Crop Leases N A g r i c u l t u r a l land use - N. Ca t t l e Farms Commercial land use - East Light Commercial N Commercial land use - East Hangar Commercial N Commercial land use - West Light Commercial N Commercial land use - West Hangar Commercial N Commercial land use - Gen. Cargo Area N Commercial land use - A i r Term. Commercial N Recreational land use - McCall Lk. Golf Course N Re s i d e n t i a l land use - C i t y of Calgary Deerfoot T r a i l N.E. N T r a n s p o r t a t i o n / u t i l c o r r i d o r s -T r a n s p o r t a t i o n / u t i l c o r r i d o r s - Main Serv Corridor N Employment - Contruction Employment N Employment - Maintenance Employment N Employment - Operations Employment N Employment - Service Employment N Regional economic a c t i v i t y - C i t y of Calgary N Adjacent landowners - N. Ca t t l e Farms N Adjacent landowners - 80th Ave. Acreages N A i r l i n e passengers - Private Passengers N A i r l i n e passengers - A i r l i n e Passengers N A i r p o r t workers - A l l A i r p o r t Personnel N R e s i d e n t i a l neighborhoods - W. Comm'ty Residents N R e s i d e n t i a l neighborhoods - S. Comm'ty Residents N R e s i d e n t i a l neighborhoods - E. Comm'ty Residents N General population - General Population N / 165 Cross Impact (cont'd) Component Name A c t i v i t y Code 11 T e r r e s t r i a l mammals T e r r e s t r i a l mammals T e r r e s t r i a l mammals T e r r e s t r i a l birds T e r r e s t r i a l mammal habitat T e r r e s t r i a l mammal habitat T e r r e s t r i a l mammal habitat T e r r e s t r i a l b i r d habitat T e r r e s t r i a l b i r d habitat Trees Trees Shrubs Grasses Crops Aquatic birds Aquatic birds Aquatic b i r d habitat Aquatic b i r d habitat Aquatic b i r d habitat Streams Streams Streams Reservoirs Reservoirs Underground aquifers S o i l s S o i l s S o i l s Airshed Airshed Airshed Airshed F i r e f i g h t i n g F i r e f i g h t i n g P o l i c e Drinking water Sanitary sewers Sanitary sewers Sanitary sewers Sanitary sewers Storm sewers Storm sewers Storm sewers Storm sewers Power l i n e s Gaslines Public transportation Public transportation Roadways Roadways Roadways Jackrabbits Badgers Coyotes Large Raptors Coyote Denning S i t e s A i r p o r t Grassed Areas Airport Crop Leases A i r Terminal B u i l d i n g N. Retent. Low Areas Barlow Rd. Orn. Groves Old Tree Nursery A.T. Landscape Shrubs Airport Grassed Areas Airport Crop Leases Waterfowl Shorebirds North Retention Pond Taxiway F Ditch F i r e Tr. Area Ditch Tank Farm Ditch East Central Ditch Cargo Area Ditch North Retention Pond S. Central Surge Pond Airport Aquifer Gulf O i l S i t e Contaminated S o i l S i t e Airport S i l t y Clays East Comm'ty Airshed West Comm' ty Airshed South Comm'ty Airshed Airport Airshed F i r e h a l l 27 (South) F i r e h a l l 13 (North) Airport RCMP 80th Ave Acreage Wells A.C. Tower Septic Pond West Hangar Sewer Line East Hangar Sewer Line A.T./Cargo Forced Mn. S.W. Storm Sewer S.C. Storm Sewer S.E. Storm Sewer N. Retent. Storm Sewer Mn Serv Corridor Power Mn Serv Corridor Gas Taxicab Services Shuttle Services Deerfoot T r a i l N.E. McCall Way McKnight Blvd. N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N M / 166 Cross Impact (cont'd) Component Name A c t i v i t y Code 11 Roadways - Barlow T r a i l N.E. N A g r i c u l t u r a l land use - Airport Crop Leases N A g r i c u l t u r a l land use - N. C a t t l e Farms Commercial land use - East Light Commercial N Commercial land use - East Hangar Commercial N Commercial land use - West Light Commercial N Commercial land use - West Hangar Commercial N Commercial land use - Gen. Cargo Area N Commercial land use - A i r Term. Commercial N Recreational land use - McCall Lk. Golf Course N R e s i d e n t i a l land use - C i t y of Calgary N T r a n s p o r t a t i o n / u t i l c o r r i d o r s - Deerfoot T r a i l N.E. Tra n s p o r t a t i o n / u t i l c o r r i d o r s - Main Serv Corridor N Employment - Contruction Employment N Employment - Maintenance Employment N Employment - Operations Employment N Employment - Service Employment N Regional economic a c t i v i t y - C i t y of Calgary N Adjacent landowners - N. Ca t t l e Farms N Adjacent landowners - 80th Ave. Acreages N A i r l i n e passengers - Private Passengers N A i r l i n e passengers - A i r l i n e Passengers N Air p o r t workers - A l l Airport Personnel N R e s i d e n t i a l neighborhoods - W. Comm'ty Residents S. Comm'ty Residents N R e s i d e n t i a l neighborhoods - N Res i d e n t i a l neighborhoods - E. Comm'ty Residents N General population - General Population N D e f i n i t i o n of Symbols: S-Significant M=Mitigable I-In- S i g n i f i c a n t U=*Unknown N=No Impact / 167 SCREENER - Summary Report Transport Canada Transports Canada Airports Authority Group Groupe de gestion des aeroports PROJECT REGISTER AND SCREENING DECISION SUMMARY Date Examined: 18/03/90 ESS Version: Ver 2.0d Region: Western Location: Calgary I n t ' l Screening Report No.: Project No.: S2P203-0034 TEC: 1542.4 F i s c a l Year: 1993/94 Project T i t l e : Demonstration Screening Project Project D e s c r i p t i o n : Northwards Extension of Main Runway Summary of Impacts Component S M I U N Component S M I U N T e r r e s t r i a l Animals Resource Harvests T e r r e s t r i a l Habitat J Recreation Vegetation V Community F a c i l i t i e s Aquatic Animals and Services Aquatic Habitat J Land Use J Surface Waters * Employment and Economy Groundwaters People J Landforms/Terrain S o i l s Atmosphere j D e f i n i t i o n s f o r symbols: S-S i g n i f i c a n t M-Mitigable I«Insignificant U»Unlcnown N-No Impact ^-conclusion based on system rules *-conclusion based on user's subjective judgement Summary of O v e r a l l Project Screening Decision An i n i t i a l assessment of the p r o j e c t leads to the fol l o w i n g conclusion: Code Decision 5 - Further study (IEE) may be required; s i g n i f i c a n t adverse impacts Prepared by: - — ^ — - — ^ — — — — — — — — — — — — — — — Environmental O f f i c e r Signature Date Approved by: - — — — — — ^ — — — Name and Designations Signature Date 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0098478/manifest

Comment

Related Items