OBJECT-ORIENTED ACTIVITY-BASED PROCESS MODELLING by DAVID LESLIE HOOD BISM., St. Francis Xavier University, 2001 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE IN BUSINESS ADMINISTRATION in THE F A C U L T Y OF G R A D U A T E STUDIES M A N A G E M E N T INFORMATION SYSTEMS THE UNIVERSITY OF BRITISH COLUMBIA February 2005 , . . . © David Leslie Hood, 2005 Abstract Many organizations seek to take advantage of the opportunities new technologies offer in the current information age. New technologies can have an impact on the fundamental organizational operations, altering well established policies and procedures. New technologies have also been poorly implemented more often than not thus mitigating the opportunities they offer. The poor implementation tends to be a direct result of a lack of proper documentation of organizational processes. This improper documentation means that any system built to support the processes that is based upon the improper documentation will itself be faulty. This thesis develops the OBPM algorithm into an objected graphical modeling language and process. The Activity-based Process Modeling (ABPM) constructs have specific and well-defined semantics for real world business process representation. Further, the change propagation algorithm which is based upon a set of ontologically derived rules is refined to create a systematic process for modeling a business process. The strength of the algorithm is from its ontological real world foundations rather than programming or data design rules of thumb. This thesis also explores the relationship of ABPM and OOEM. Both languages are designed to model a specific view of organizational activity, irrespective of how a later information system artifact will be built. By relating the two grammars using ontological foundations we can acquire greater understanding of an organization without losing information. Finally, this thesis proposes a set of design principles for an ABPM CASE tool that is implementation independent which means that no matter how one decides to implement ABPM if they follow our requirements they will be able to create a tool to fully support the business process model generation process. u TABLE OF CONTENTS ABSTRACT ii T A B L E OF CONTENTS » iii LIST OF FIGURES viii LIST OF TABLES xii ACKNOWLEDGEMENTS xiv 1. INTRODUCTION 1 1.1. Motivation 1 1.2. Thesis Objectives 4 1.3. Thesis Outline 5 2. PROCESS MODELING INTRODUCTION..... 7 i ^ i . . . : . v i . 2.1.. Introduction 7 ' , \ . ( . . . . . 2.2. Definition of Business Process Modeling 8 2.3. The'Terminology 9 2.4. The Ideal Process Model 10 2.5. Ontology 11 2.5.J. BWWP Ontology 14 2.6. Current Business Process Modeling Techniques 14 2.6.1. Colored Petri Nets , 15 2.6.2. Integrated Definition 3 17 2.6.3. Event-controlled Process Chains 19 2.7. Summary ' 20 O B J E C T - O R I E N T E D A C T I V I T Y - B A S E D P R O C E S S M O D E L L I N G (ABPM) 22 3.1. Introduction "-22 3.2. Ontology-Based Process Modeling (OBPM) 22 3.2.1. OBPM Algorithm 24 3.3. Object-Oriented Activity-Based Process Modeling grammar 25 3.3.1. Classes and Objects 26 3.3.2. Association 28 3.3.3. Object Communication 29 3.3.4. The Combination of OBPM and Object-Orientation 30 3.3.4.1. Attributes 31 3.3.4.2.Operations and Change Propagation 31 3.3.4.3. Activities ". 32 3.3.4.4. Laws 33 3.3.5. Mapping Summary 34 3.4. Meta-model of ABPM 34 3.5. Graphical Representation 37 3.5.1. Domain Representation 38 3.5.2. External Agent 38 3.5.3. Internal Agent 39 3.5.4.. Resources 42 3.5.5. Agents Sharing A Mutual Attribute 42 3.5.6. Composite Agents 44 3.5.7. Superagents and Subagents 47 3.5.8. An Alternate Notation 48 3.5.9. Allowed Interactions 51 3.6. A B P M Modeling Process 52 3.6. J. Modeling Rules 53 3.6.2. Modeling Process 57 3.6.3. Model Integrity 59 3.1. An Example • 60 3.8. Summary 64 4. L I N K I N G O O E M A N D A B P M 65 4.1. Introduction 65 4.2. Object-Oriented Enterprise Modeling (OOEM) 66 4.2.1. OOEM constructs 66 4.2.2. Request Propagation 67 4.2.3. How To Generate An OOEM. 67 4.2.4. Graphical Representation Of OOEM. 68 4.2.5. A Shortcoming of OOEM. 69 4.3. Basis O f Conversion 69 ' - \ n 4.4. A B P M To O O E M Conversion 75 4.4.1. A B P M To O O E M Conversion Steps 75 4.4.2. A B P M To O O E M Conversion Step Derivation 75 4.5. O O E M To A B P M Conversion 81 4.5.1. O O E M To A B P M Conversion Steps 81 V 4.5.2. O O E M T O A B P M Conversion Step Derivations 82 4.6. Converting Decompositions 89 4.7. Converting Subagents/Superagents And Subclasses/Superclasses 89 4.8. Summary 89 5. DESIGN PRINCIPLES OF A N ABPM CASE TOOL 90 5.1. Introduction 90 5.2. System Goals 90 5.3. System Requirements 90 5.4. Constructs To Represent 91 5.5. User Interactions 95 5.5.1. Drawing A B P M Diagram Interactions 95 5.5.1.1. Internal and External Agents 95 5.5.1.2. Agents Sharing A Mutual Attribute 95 5.5.1.3. Superagents and Subagents 96 5.5.1.4. Composite and Component Agents 97 5.5.2. Display Option 98 5.5.3. O O E M and A B P M Conversions 99 5.5.3.1. A B P M To O O E M Converter 99 5.5.3.2. O O E M To A B P M Converter 101 5.5.4. Semantic Checker 104 5.6. Design Limitations 106 5.7. Summary 107 6. CONCLUSIONS AND FUTURE RESEARCH 108 6.1. Thesis Summary 108 6.2. Contributions 109 6.3. Limitations A n d Future Research 110 B I B L I O G R A P H Y 113 A P P E N D I X A - S T E P B Y S T E P D E R I V A T I O N O F T H E A B P M F O R T H E A C M E W A R E H O U S E M A N A G E M E N T C A S E 116 A P P E N D I X B - A G E N T A C T I V I T Y T E M P L A T E S F O R T H E A C M E W A R E H O U S E M A N A G E M E N T C A S E 146 A P P E N D I X C - A D I S C U S S I O N O F T H E B W W O N T O L O G Y 153 C l Static M o d e l o f Things '. 153 C.2 Dynamic M o d e l o f Things 154 C.3 Static M o d e l o f Systems 155 C . 4 Dynamic Mode l o f Systems 156 A P P E N D I X D M O D E L I N G G R A M M A R E X A M P L E S 158 D. l Colored Petri Nets 159 D.2 I D E F 3 162 D.3 E P C : ..: 165 vii Lis t of Figures Figure 3-1 ABPM Agent 30 Figure 3-2 ABPM Meta-Model 36 Figure 3-3 Domain Representation 38 Figure 3-4 External Agent Representation 38 Figure 3-5 Internal Agent Representation 39 Figure 3-6 Resource Representation 42 Figure 3-7 Agents Sharing A Mutual Attribute Representation 43 Figure 3-8 Different Operations changing one incoming interface attribute 43 Figure 3-9 Composite Agent Representation 44 Figure 3-10 Adding A Composite Agent 45 Figure 3-11 Decomposition Of A Composite Agent 46 Figure 3-12 Superagent and Subagent Representation 47 Figure 3-13 Creating A Superagent 48 Figure 3-14 Example Compressing Down To Agent Interactions 48 Figure 3-15 Example Compressing Down To Activity Name 48 Figure 3-16 Internal Agent Template 49 Figure 3-17 Resource Template 49 Figure 3-18 External Agent Template 50 Figure 3-19 Composite Agent Template 50 Figure 3-20 Subagent Agent Template 51 Figure 3-21 Allowed Interactions For An External Agent 51 Figure 3-22 Allowed Interactions For A Resource 51 viii Figure 3-23 Allowed Interactions For An Internal Agent 52 Figure 3-24 Compressed A B P M Showing Agent Interactions 62 Figure 3-20 A B P M For The A C M E Warehouse Management Case 63 Figure 4-1 Proposed Relationship And Its Foundations 65 Figure 4-2 O O E M For The A C M E Warehouse Management Case 68 Figure 4-3 The Enterprise Modeling Approach 70 Figure 4-4 O O E M reinterpreted as a state change view 70 Figure 4-5 The Process Modeling Approach 71 Figure 4-6 A B P M reinterpreted as a state change view 71 Figure 4-7 The Ontological Meaning Approach 72 Figure 4-8 Demonstrating Step 1 External Object Conversion; A B P M To O O E M . 76 Figure 4-9 Demonstrating Step 2 Internal Object Conversion; A B P M To O O E M : 77 Figure 4-10 Demonstrating Step 3 Request and Response Identification; A B P M To O O E M 78 Figure 4-11 Demonstrating Step 4 Interface Attribute Conversion; A B P M To O O E M 79 Figure 4-12 Demonstrating Step 5 Internal Attribute Conversion; A B P M To O O E M 79 Figure 4-13 Demonstrating Step 6 Service Creation; A B P M To O O E M 80 Figure 4-14 Demonstrating Step 1 External Agent Conversion; O O E M To A B P M 82 Figure 4-15 Demonstrating Step 2 Internal Agent Conversion; OOEM To ABPM 83 Figure 4-16 Demonstrating Step 3 Outgoing Interface Attribute Identification; OOEM To ABPM 84 Figure 4-17 Demonstrating Step 4 Incoming Interface Attribute Identification; OOEM To ABPM 84 Figure 4-18 Demonstrating Step 5 Activity Creation; OOEM To ABPM 85 Figure 4-19 Demonstrating Step 6; Outgoing Interface Attribute Assignment; OOEM To ABPM . 86 Figure 4-20 Demonstrating Step 7 Internal Attribute Identification; OOEM To ABPM 87 Figure 4-21 Demonstrating Step 8 Operation Identification; OOEM To ABPM..88 Figure 4-22 Demonstrating Step 9 Resource Identification; OOEM To ABPM...88 Figure 5-1 The OOEM Metamodel 92 Figure 5-2 OOEM metamodel with ontological meaning included 93 Figure 5-3 ABPM metamodel with ontological meaning included 94 Figure A- l Office Clerk Is Stable 122 Figure A-2 After Step 3 For Order Status::Customer And Withdrawal Request:: Warehouse 124 Figure A-3 A Tree Showing All Of The Changes 140 Figure A-4 Solution to the ACME Warehouse Case 141 Figure A-5 Creating An Employee Superagent 143 Figure A-6 The Decomposition Of Warehouse 144 x Figure A-7 Warehouse Is A Composite Agent 145 Figure B-l Compressed Agent Interaction Diagram 146 Figure B-2 Forklift Operator Agent Template 146 Figure B-3 Office Clerk Agent Template 147 Figure B-4 Truck Driver Agent Template 147 Figure B-5 Planner Agent Template 148 Figure B-6 Customer Agent Template 148 Figure B-7 Employee Agent Template 148 Figure B-8 Warehouse Agent Template 149 Figure B-9 Warehouse Agent Template in 'Composited' view 149-150 Figure B-10 Warehouse Worker Agent Template 151 Figure B-l 1 Warehouse Manager Agent Template 152 Figure D-l Colored Petri Net for the ACME Warehouse Case 160 Figure D-2 Process Flow Description for the ACME Warehouse Case 163 Figure D-3 Object State Transition Network for the ACME Warehouse Case... 164 Figure D-4 EPC for the ACME Warehouse Case 167-168 List of Tables Table 1 Relative Cost To Fix A n Error 3 Table 2-1 Effects O f A Redesigned Process 8 Table 2-2 B W W Ontology Constructs 13 Table 2-3 B W W P Ontological Analysis o f Colored Petri Nets 16 Table 2-4 B W W P Ontological Analysis o f I D E F 3 18 Table 2-6 B W W P Ontological Analysis o f E P C 20 Table 2-6 Demonstrating M O O and M O C 21 Table 3-1 Mapping Ontological Constructs A n d Premises To Object-Oriented Constructs A n d Premises 26 Table 3-2 Mapping Summary 35 Table 3-3 Relating The Rules To The A B P M A n d O B P M Algori thms 59 Table 4-1 Conversion Table 73 Table 4-2 Questions o f interest in an A B P M to O O E M Conversion 75 Table 4-3 Questions o f interest in an O O E M to A B P M Conversion 81 Table A - l Step 3 For Order Status::Customer 123 Table A - 2 Step 3 For Withdrawal Request::Warehouse Manager 123 Table A - 3 Step 3 For Item Existence::Warehouse 124 Table A - 4 Step 3 For Search Results::Warehouse Manager 125 Table A - 5 Step 3 For Transport Form::Planner 126 Table A - 6 Step 3 For Transport Schedule::Warehouse Manager 127 Table A - 7 Step 3 For Order Status::Office Clerk 128 Table A - 8 Step 3 For Transport Orders: :Truck Driver 129 x i i Table A-9 Step 3 For Pickup Notification::Warehouse Manager 130 Table A-10 Step 3 For Ready Items::Warehouse Worker 131 Table A- l 1 Step 3 For Move ltems::Forklift Operator 131 Table A-12 Step 3 For Truck Status::Planner 132 Table A-13 Step 3 For Ready To Load::Warehouse Worker 133 Table A-14 Step 3 For Loaded::Truck Driver 134 Table A-15 Step 3 For Dropoff Notification::Warehouse Manager 135 Table A-16 Step 3 For Ready Receiving::Warehouse Worker 136 Table A-17 Step 3 For Ready To Unload::Warehouse Worker 137 Table A-18 Step 3 For Update Form::Warehouse Manager..... 138 Table A-19 Step 3 For Arrival::Warehouse 139 Table A-20 Step 3 For Items And Documentation::Customer 139 Table D-l Declarations for the ACME Warehouse Colored Petri Net 161 xiii Acknowledgements This thesis would not have been possible without the assistance and guidance of Professor Yair Wand, my thesis supervisor. I am also indebted to Professor Carson Woo for his many insightful and helpful discussions. I also wish to extend thanks to two incredible people Young Eun Lee and Hang (Jasmin) Zeng. Without the emotional and personal support of the two of you I would never have made it this far. Thank you all. 1. Introduction Documenting the processes of an organization is not a new concept. Organizations document what is occurring in a process in an attempt to capture what happens in the organization. Process models can be used to show where time and money are being spent and can identify things like inefficiencies and bottlenecks. One of the more common uses of process documentation is for business process reengineering (BPR). Organizations that can successfully reengineer their processes often achieve a large significant competitive advantage over their rivals. BPR typically involves a technological integration that was heretofore unseen. However implementing technology is a poorly performed function. Typically information system projects run over budget, over time, or the output does not perform what is required. Typically, the problems that occur can be traced back to incomplete or unclear requirements concerning what occurs in a process (Standish, 1994, CIO, 1997). 1.1 Motivation In the field of MIS there exist many process modeling methods, including EPC, IDEFO, IDEF3, U M L Activity Diagrams, EDPM, Petri Nets, and PERT/CPM to name a few. Of the methods not one has emerged as the dominant paradigm for organizational process modeling. This raises the rather simple question of "why not?" Looking at the literature about process modeling in the MIS field, we see that there is a growing trend to suggest using several process modeling languages to capture all the relevant information. An example of this would be the work of Bosilj-VukSic & Hlupic 1 (2001), in which they suggest using both Petri Nets and IDEF3 for business process modeling. The combination of multiple methods suggests that the methods themselves are deficient and cannot fully represent a business process fully on their own. This apparently answers our question above about why there is not a dominant process modeling methodology, however this raises a new question of "Is there a way to represent a process completely in one modeling methodology?" Before we answer this question we also need to consider why it is important to create better business process models. An information system is an artifact that represents another 'real world' system. If we can create a better model of the real world system then we should be able to create better artifacts that represent that system. Developing an information system has been defined as a three step transformation of Analysis, Design, and Implementation1 (Wand and Weber, 1990, p. 125). The analysis stage is where what an information system artifact will do is formalized based upon some process model, i.e. a representation of what will happen is created. The design stage is where the requirements (i.e. the representation) are translated into a model of the information system. The Implementation stage is where the information system is implemented based on the output of the design stage. Everything is built up from the analysis stage, therefore if there are errors made in the analysis stage they will propagate through to the information system itself (Grause and Weinberg, 1989, Wand and Weber, 1995). Thus, better representations lead to better systems. 1 We acknowledge'there could be others, such as development or testing, but for the purposes of this section the three phases are sufficient. 2 In terms of cost to fix an error the earlier an error is found in system development the cheaper it will be. If we consider the cost ratio the difference is staggering. This difference is also sensible since it is less expensive to, for example, draw a new diagram than it is to build a new system. Phase in Which Found Cost Ratio Requirements 1 Design 3-6 Coding 10 Development Testing 15-40 Acceptance Testing 30-70 Operation 40-1000 Table 1. Relative Cost to Fix an Error (Grause and Weinberg, 1989) In our work on conceptual modeling we will be focusing above the requirements stage of system development. In theory this should have an even lower cost ratio if errors are made and found here than 1. Hence if we can improve process models and catch more errors there will also be an impact on the bottom line of an organization. There is a means of fully representing things that has been successfully used in the MIS field. Ontology, a philosophy concerning what exists, has been applied to the area of enterprise modeling and serves as the basis of object-oriented enterprise modeling (OOEM), however it had not been applied to process modeling until recently (see, for example, Wand, Woo, & Jung, 2000, Wand & Woo, 2002 Zhao, 1995). In an attempt to assist in developing a robust organizational process modeling methodology to solve the problems mentioned above Wang (2002), proposed Ontology-based process modeling (OBPM). Currently OBPM exists only as an algorithm. 3 1.2 Thesis Objectives As mentioned, O B P M exists only as an algorithm. This makes both using O B P M difficult and the output difficult to interpret. To address this deficiency we will present a graphical modeling grammar. This notation will be from the perspective of object orientation. Thus the concepts of encapsulation, composition, classification, and communication will be considered. A graphical notation unto itself is not necessarily useful i f it is not clear how to use the notation. Thus following the notation there must be a process for generating an Object-Oriented O B P M . The algorithm presented in Wang (2002) is textual and was not designed with graphics or object-oriented constructs in mind. Hence we wil l develop a process for generating an Object-Oriented O B P M . Another point of interest would be that since O O E M and O B P M are both based on the Bunge-Wand-Weber (BWW) Ontology and are used to model different organizational perspectives then they should be related. We will develop a means to convert from the activity-abased Object-Oriented O B P M view to the interacting agent-based O O E M view using the B W W P ontology as the basis of conversion. Further, a C A S E tool for O O E M has been developed. It stands to reason, since there is a relationship between O O E M and O B P M , and there exists a C A S E tool for O O E M , that it is possible to develop a C A S E tool for O B P M . This also leads to the conclusion that i f they can both be represented in a C A S E tool, then conversion between O O E M and O B P M can also be captured in a C A S E tool. Therefore a set of design principles for an Objected-Oriented Ontology-Based Process Modeling C A S E tool will be developed. The objectives of this thesis can be summarized as follows: 1. Formalize O B P M into an object-oriented graphical modeling grammar. 2. Present a process for generating object-oriented O B P M diagrams 3. Explore and discover the relationship between O O E M models and Object-oriented O B P M models and present a means of conversion between them using the B WWP ontology as a basis of conversion. 4. To present a set of design principles for an Object-Oriented O B P M C A S E tool and link it to the architecture of an O O E M C A S E tool. 1.3 Thesis Outline This thesis consists of five chapters. Chapter two provides an introduction to process modeling and the related terminology. It also presents the different aspects of process modeling. Examples of process modeling techniques are presented to illustrate the different process modeling aspects, and to illustrate flaws present in ubiquitous methodologies. Chapter three concerns Ontology-based Process Modeling (OBPM). It formalizes O B P M into an object-oriented modeling grammar called object-oriented activity based process modeling(ABPM). It then introduces graphical modeling grammar for representing 5 object-oriented activity-based process models. It also presents a process for how to create an ABPM. The chapter includes an example to illustrate these points. i Chapter four introduces Object-oriented Enterprise modeling (OOEM). Since the foundations of OOEM and ABPM are similar, this chapter links the two methodologies. It also provides steps relating to how to switch between an ABPM and an OOEM and vice versa. It will include a running example to demonstrate the results. Chapter five is a look at the design principles for the implementation of ABPM as a CASE tool. It discusses what to add to an existing OOEM CASE tool to both represent ABPMs and for conversion between OOEMs and ABPM. It ends with a discussion of the limitations of the proposed design. Chapter six concludes the thesis with a summary and discussion of its contributions. It also suggests further research that can be done. 6 2. P r o c e s s M o d e l i n g I n t r o d u c t i o n 2.1 Introduction Information systems offer the potential to supplement the ability to increase revenues while lowering costs in the value creation process of a business by greatly improving both the efficiency and the effectiveness of a process. This is where business process modeling comes in to play in the information systems field. From the business process engineering perspective a process model can serve as: a focus for discussion, a way of communicating a process to others, a starting point for analysis, a starting point for design, a baseline for monitoring process improvement, and control for a real world process (Huckvale and Ould, 1994). Before we present the object-oriented ontology-based process modeling grammar, we will provide a discussion of the basic concepts behind business process modeling. We will also discuss the basic terminology used in business process modeling. We will then look at what has been proposed for the ideal process model. We then delve into the ontological perspective, what it is, and why it is a better perspective than other views. We will then look at several business process modeling languages. These languages include Petri Nets, Integrated Definition 3 (IDEF3), and Event-controlled Process Chains (EPC). The languages are some of the most ubiquitous process modeling techniques. These languages will be analyzed in terms of their (process-oriented) ontological completeness and (process-oriented) ontological clarity. Our goal is to demonstrate how the languages are deficient and why object-oriented ontology based process modeling will be beneficial. 7 2.2 Definition of Business Process modeling All businesses strive to achieve their goals via some sort of process. A business process can be defined as the sequence of activities that lead to value creation for the consumer of a business' goods and services. If a good or service is perceived to have a higher value then consumers should be willing to pay more for the good or service. That said it stands to reason that businesses will want to analyze their processes, with the goal of improving them, so that their profits will be increased. It has been argued (e.g. Hui, 1997, Romney, 1994, VBM, 2004) that redesigning a business process can have many impacts on and benefits for an organization. A brief non-exhaustive list of the impacts and benefits is presented below in table 2-1. Impact of redesigned process Benefits of the impacts 1. Task elimination ' 2. Bottleneck and delay elimination 3. Parallel work enabled 4. Eliminate rework and redundancy 5. Decreased defects 6. Eliminate staff 7. Less time spent on non value-added activities 1. Improved productivity 2. Reduced cycle times 3. Reduced costs ; 4. Improved customer service 5. Improved quality and consistency 6. Increased revenues/charging higher prices 7. Increased competitiveness 8. Improved forecasting 9. Better capacity utilization 10. Quicker delivery of new products and services 11. Greater workload capacity Table 2-1 Effects of a redesigned process (partially adapted from, Hui, 1997, p.8) Before we define business process modeling we need to look at and consider modeling itself. A model is an abstraction from the real world of something in the real world. In this abstraction we leave out the facets of whatever we are modeling that are not relevant, and emphasize those that are pertinent. If we combine this definition of modeling with our above definition of a business process, we can define business process modeling as: the creation of an abstract representation of the real or proposed sequence of value creation activities performed within an organization in order to achieve its business goals. Adapting from Wang (Wang, 2002), the business process model will therefore include: participants in the business process, the events that start activities in the business process, the activities triggered by the events in the business process, the sequence of the activities of the business process that will be performed, the sequence that events in a business process occurred, the resources that are utilized in an activity of a business process, and the inputs and outputs of each activity of the business process. 2.3 The Terminology In an analysis of process modeling methodologies, Wang (Wang, 2002) found that traditionally, the following constructs have been used in business process modeling: process, agent, non-agent or resource, activity, operation, event, state, data, logical connector, business rule, input, and output. However, Wang (Wang, 2002), points out that many languages either do not include all of these constructs (construct incompleteness), use many elements of the modeling language to represent the same construct (construct redundancy), or use the same 9 construct to represent many elements (construct overload). This causes process models using these languages to be incomplete and/or ambiguous. 2.4 The Ideal Process Model Various process modeling methodologies focus on different aspects of what is being modeled (Wang, 2002). The majority of process modeling languages fall into four categories (Huckvale and Ould, 1994, citing Curtis et al., 1992): 1. Functional. A functional process modeling language is concerned with representing what activities are being performed and the dataflows that connect them. Examples in this category include EPC (Event-controlled Process Chains), IDEF3 (Integrated DEFinition 3), and EDPM (Event-Driven Process Modeling). 2. Behavioural. A behavioural process modeling language is concerned with representing when activities occur. They use sequencing, feedback loops, iteration, decision making, triggers, and the like. Examples in this category include PERT/CPM (Project Evaluation Review Technique/Critical Path Method), Petri Nets, and EPC. 3. Organizational. An organizational process modeling language is concerned with representing where and by whom activities are performed. They also include the physical communication mediums and storage media. Examples in this category include EPC, and EDPM. 4. Informational. An informational process modeling language is concerned with representing the data entities that are generated or manipulated by a process/ such as documents, data, artifacts, and products. This includes their structure and 10 interrelationships. Examples of this include DFD (dataflow diagrams), IDEF3 and E D P M . Due to the fact that these four categories cover different areas, using a methodology from each category should cover all the information necessary to fully represent a process. However (as noted by Wand & Weber, 2002) this leads to potential issues of redundancy and sufficient coverage. Redundancy refers to the same construct being represented more than once. Sufficient coverage refers to all the relevant items of interest being captured. This leads to the premise i f a modeling language included all of the constructs from section 2.3 it would thus cover the functional, behavioural, organizational, and informational perspectives2. This is one of the bases for Ontology-based Process Modeling (OBPM). O B P M will be discussed in chapter 3. 2.5 Ontology Ontology is a branch of philosophy about what exists in the real world. It is of prime concern when it comes to information systems modeling since the model should accurately reflect what does or will exist in the real world 3. If the model does not accurately reflect this, then anything created based on the model (in this case an information system) will itself be deficient. 2 Data itself would be left out of the ideal model since as Huckvale and Ould (1994) note data is typically used to record the state of a process due to people having poor memories or as a means of implementing a transaction. The transaction,itself is important not how it is implemented at this level. 11 According to Wand and Weber (Wand and Weber 1993, 1995) for a grammar to be ontologically expressive, it needs to be both ontologically complete and ontologically clear. Ontological completeness refers to there being at least one construct in the grammar for every ontological construct. If this does not occur then construct incompleteness (or construct deficit) occurs. Ontological clarity refers to how clearly an ontological construct is represented in a grammar. Ontological clarity can be undermined via construct overload, construct redundancy, and construct excess. Construct overload occurs when an element of the grammar represents more than one ontological construct. Construct redundancy occurs when more than one element of a grammar represents a single ontological construct. Construct excess occurs when there is a construct in the grammar that does not represent any ontological construct. Typically when a modeling grammar is ontologically deficient predictions can be made as to where the language will suffer in capturing information about the domain of interest. Approaches to solve the deficiencies usually involve using more than one grammar to model what is going on. However there is no formal basis for selection of which grammars to use other than rules of thumb or things seeming to fit and workout. 3 In this case what will exist is not definite, but rather refers to a proposed information system that may or not be built. 12 Object-Oriented Enterprise Modeling and Ontology-Based Process Modeling are both based on what has come to be known as the Bunge-Wand-Weber (BWW) Ontology. The main principles of the BWW ontology can be summarized as follows (Wand and Woo, 1999): • The world is comprised of things • Things possess properties • Things interact • Every thing changes and every change changes things • Things with similar properties can be grouped into classes The constructs of BWW Ontology can be organized into four categories as shown in table 2-2 below (Zhao, 1995, Wang 2002). A complete discussion of the B W W constructs can be found in appendix C. Category BWW Ontology Construct Static model of a substantial individual 1. Thing, composite thing 2. Property, intrinsic or mutual property, hereditary or emergent property4 3. State, conceivable state space, state law 4. Class, kind, natural kind Dynamic model of a substantial individual 1. Event, conceivable event space 2. Transformation, transformation law 3. History Static model of a system 1. Coupling 2. System 3. System Composition 4. system environment 5. system structure 6. system decomposition 7. subsystem 8. level structure Dynamic model of a system 1. Stable and unstable state 2. Internal event and external event 3. Well defined event and poorly defined event Table 2-2 BWW Ontology Constructs 4 Here the use of'or' does not mean intrinsic is interchangeable with mutual and hereditary is interchangeable with mutual. 'Or' in this sense is a restriction that the property can be one of them but not both. 13 2.5.1 BWWP Ontology Some of the BWW ontology concepts are closely related to the concepts of process. The process related BWW ontology concepts have been adapted to create the BWWP (BWW Process) ontology. The BWWP constructs include the BWW ontological constructs of: thing (simple thing and composite thing), property (intrinsic property and mutual property, hereditary property and emergent property), state (stable state and unstable state), event (internal event and external event), transformation, and law (state law and transformation law). The new constructs added in BWWP are: actor, non-actor, actuator, propagator, and process (Wang 2002). The new constructs of BWWP are explained as follows. A thing can be either an actor or a non-actor. A non-actor does not change the state of any thing including the non-actor itself. Only an actor can change the state of a non-actor. An actor is an actuator if it changes the state of at least one other thing (actor or non-actor). An actuator is a propagator if it changes the state of at least one other actor. A process is a set of actors and non-actors that interact with one another. A process is activated one or more actors is changed from a stable state to an unstable said. A process can be triggered by one or more events. When all actors and non-actors are in stable states a process is completed (Wang, 2002). 2.6 Current Business Process Modeling Techniques The purpose of this section is to demonstrate the differences between modeling methodologies and highlight their ontological deficiencies. The methodologies we have 14 chosen include Colored Petri Nets, Integrated Definition (IDEF3), and Event-controlled Process and Chains (EPC). Petri Nets and IDEF3 were chosen due to the aforementioned suggestion of using them together. EPC was chosen due to the fact it was developed and is used by one of the largest providers of ERP providers SAP. It is beyond the scope of this paper to perform an ontological analysis on all business process modeling techniques In order to perform this comparison we require a common reference point. The modeling languages will be analyzed using the BWWP ontology. For those unfamiliar with the grammars, an example using each grammar is provided in Appendix D. 2.6.1 Colored Petri Nets Table 2-3 maps the Colored Petri Net constructs to BWWP Ontological Constructs. It is readily apparent that Colored Petri Nets suffers from many ontological deficiencies. First of all, construct deficit is present with respect to properties. A'token is possessed by a place, making it a property of a place. However there is no concept of an intrinsic token possessed naturally by a place, any token can leave or enter a place based on the firing rules established. There is also no concept of an intrinsic arc or intrinsic firing rule. Other places and transitions need to be establishedto have arcs and firing rules. Since there are no intrinsic properties, there can be no hereditary or emergent properties. The second ontological deficiency Petri Nets suffers from is construct redundancy. Consider the BWWP construct of a property. Where there is a mapping, there is overload between arcs and tokens. Is a property a token, or one of the variations on a directed arc? 15 BWWP Construct Colored Petri Nets Construct Thing Place Simple Thing. . _ Place - . . Composite Thing Place Actor Place Non-actor Place Actuator Place Propagator Place Property Token, Directed Arc, Inhibitor Arc, Clearing Arc, Priority Transition Intrinsic property N/A ' Mutual property Token, Directed Arc, Inhibitor Arc, Clearing Arc, Priority Transition Hereditary property N/A Emergent property N/A State Distribution of Tokens Stable state No transitions are enabled Unstable state One or more transitions are enabled Event Transition Internal event Transition External event Transition Transformation f' Transition Law Directed Arc, Inhibitor Arc, Clearing Arc, Priority Transition State law i.'''' Directed Arc, Inhibitor Arc, Clearing Arc, Priority Transition Transformation law Directed Arc, Inhibitor Arc, Clearing Arc, Priority Transition Process Colored Petri Net Table 2-3 BWWP Ontological Analysis of Colored Petri Nets Finally Petri Nets also suffers from construct overload. Arcs can represent both properties and laws. The construct of a transition is also overloaded. Is a transition a transformation or an event? , ) Since it suffers from construct deficit, construct redundancy, and construct overload Petri Nets are.both ontologically incomplete and ontologically unclear. 16 2.6.2 Integrated Definition 3 IDEF3 has two main components for modeling: process flow description and object state transition networks. Table 2-4 maps the IDEF3 constructs to BWWP ontological constructs. We start the analysis of IDEF3 with the process flow description diagram. Process flow description diagrams suffer from construct deficit, construct overload, construct redundancy, and construct excess. Construct deficit is readily apparent from the complete lack of a construct to represent a thing. Also, properties, states, and events are only partially represented. The UOB, links, junctions, and referents are all overloaded. That is, they are used to represent more than one ontological construct. Construct redundancy occurs with property, law, and process. Lastly, the excess constructs occurs with decomposed UOB, UOB numbering, link numbering, junction numbering, partial descriptions, elaborations, and notes. Thus process flow description diagrams are neither ontologically complete nor ontologically clear. Object state transition networks suffer from construct deficit, construct overload, construct redundancy, and construct excess. Construct deficit exist since there is no representation of a property and only partial representations of thing and event. Construct overload is present since a referent can be used to represent both a transformation and an event. Construct redundancy occurs with state, event, law, and process. Construct excess exists since UOB numbering, link numbering, junction numbering, partial descriptions, elaborations, and notes have no ontological meaning. 17 BWWP Construct Process Flow Description Object State Transition Networks Enhanced Transition Schematics Thing Object Object, first order objects, second order objects Simple Thing Object kind symbols Object kind symbols Composite Thing Hiding Object State information Hiding Object State information Actor Individual symbol Non-actor Individual symbol Actuator Propagator Property Links,Junctions,referents Links,Junctions,referents Links,Junctions,referents Intrinsic property Mutual property Links Relation Hereditary property — Emergent property State Activation plots Object state symbols, interval diagram Object state symbols, interval diagram Stable state State conditions, Exit Conditions State conditions, Exit Conditions Unstable state State Conditions, Exit Conditions, •. State Conditions, Exit Conditions •• Event UOB . UOB, scenario, Referent • UOB, scenario Referent Internal event External event Transformation UOB Link, referent Link, referent Law Links, junctions, referents Link, junction, referents Link, junction, referents State law Simple precedence links, constrained precedence links, dashed links, junctions Weak transition link, strong transition link, call and continue referent Weak transition link, strong transition link, call and continue referent Transformation law Simple precedence links, constrained precedence links, dashed links, junctions Entry Conditions, Transition Condition, referent, junctions call and wait referent, referents attached to the same point, temporal indeterminancy marker Entry Conditions, Transition Condition, referent, junctions call and wait referent, referents attached to the same point, temporal indeterminancy marker process ' Scenario, activation plots Scenario, Object schematics, complex transition schematic Scenario, Object schematics, complex transition schematic Decomposed UOB, UOB numbering, link numbering, junction numbering, partial descriptions, elaborations, notes UOB numbering, link numbering, junction numbering, partial descriptions, elaborations, notes UOB numbering, link numbering, junction numbering, partial. descriptions, elaborations, notes Table 2-4 BWWP Ontological Analysis of IDEF3 18 Enhanced transition schematics suffer from the same problems as object state transition networks, except the construct deficit is lower. Enhanced transition schematics partially represent properties, and better represent a thing than object state transition networks. Since the IDEF3 diagrams are meant to be used together, we need to consider whether they achieve minimum ontological overlap (MOO) and maximum ontological coverage (MOC). MOO is when an ontological construct is represented in only one diagram with the goal of reducing representations of a domain that are in conflict. MOC is when in the combination of diagrams covers all phenomena with the goal of having a complete representation of the domain (Wand and Weber, 2002, citing Green 1996). MOO is not achieved in IDEF3. An example of the lack of MOO is process flow descriptions use activation plots to represent states whereas the object transition diagrams use object state symbols and interval diagrams. MOC is not achieved is IDEF3. An example of the lack of MOC is that none of the diagrams have a construct to represent an intrinsic property. 2.6.3 Event-controlled Process Chains Table 2-5 Maps the EPC constructs into ontological constructs To begin, EPC suffers from construct deficit. The concepts of Property, Thing, and Event are only partially represented. Since things are what perform actions, are affected, etc. and things are not fully represented we cannot fully represent what performs actions, are affected, etc. When we look at event it is not immediately clear about those that are internal or external to the domain of interest. It can be argued that the event that starts the process chain is external, but how is it known which events generated in the domain affectsomething outside the domain. B W W P C o n s t r u c t E P C C o n s t r u c t Thing Organizational Unit Simple Thing Composite Thing Actor Organizational Unit Non-actor Actuator propagator Property Intrinsic property Information/Material flow Mutual property Hereditary property Emergent property State Stable state No Events occur Unstable state Event Event Event Internal event: External event Transformation Task Law Control Flow + Connectors State law Control Flow + Connectors Transformation law Control Flow + Connectors Process Process Information Object, Organization assignment Table 2-5 BWWP Ontological Analysis of EPC To end EPC also suffers from construct excess since the information object construct and organization assignment constructs have no BWWP ontological meaning. Since it suffers from construct deficit, and construct excess EPC is both ontologically incomplete and ontologically unclear. 2.7 Summary If we just do a simple count of the number of constructs each grammar has (see table 2-6) our three grammars never achieve MOC even when all three are used together. 20 I D E F 3 BWWP Construct Colored Petri Nets Process .Flow Description Object State Transition Networks Enhanced Transition Schematics Event-Driven Process Chains Total Thing 2 0 1 3 1 7 Simple Thing 2 0 1 1 0 4 Composite Thing 2 0 1 1 0 4 Actor 2 0 0 1 1 4 Non-actor 2 0 0 1 0 3 Actuator 2 0 0 0 0 2 propagator 2 0 0 0 0 2 Property 5 3 3 3 0 11 Intrinsic property 0 0 0 0 1 1 Mutual property 5 1 0 1 0 7 Hereditary property 0 0 0 1 0 1 Emergent property 0 0 0 0 0 0 State 1 0 2 2 0 5 Stable state 1 0 2 2 1 6 Unstable state 1 0 2 2 1 6 Event 1 3 3 1 9 Internal event 1 •" •'' • • 0 0 o ' '• 0 ' 1 External event 1 • 0- ' ' 0 • o 0 1 Transformation 1 1 2 - 2 1 • : . . 7 Law 1 3 3 3 2 12 State law 5 4 3 3 2 18 Transformation law 5 4 7 7 2 25 Process 5 2 3 3 1 14 No mapping 0 7 6 6 1 20 Table 2-6 Demonstrating MOO and MOC Also the multiple grammars really begin to make problems with M O O grow. To solve this problem we suggest using one grammar designed using the B W W P ontology namely ontology based;process modeling (OBPM). O B P M is discussed in the next chapter. 21 3. Object-Oriented Activity-Based Process Modeling (ABPM) 3.1 " Introduction _ . . . _ . The purpose of this chapter is threefold. The first goal is to introduce Ontology-Based Process Modeling (OBPM). The second goal is to present a modeling grammar that can be used to generate object-oriented activity-based process modeling diagrams. The third goal is to introduce a modeling process to guide in the construction of ABPM diagrams in a systematic way. This chapter closes with an example of an ABPM diagram based on the ACME Warehouse Management case. In order to accomplish the goals of this chapter the following concepts will be introduced: agent, activity, operation, attribute, resource, and law. The relationships between the concepts will also be developed. 3.2 Ontology-Based Process Modeling (OBPM) After having identified so many deficiencies with other ubiquitous process modeling methodologies the question of is there a better method to capture business processes remains? The answer is yes and no. Ontology-based process modeling (OBPM) has been proposed. OBPM attempts to capture what things are involved in a process, what they do, and what is done to them. It does have the advantages of being real world rather than information systems oriented which allows for representations of a process independent of how it will be implemented, ontologically complete and clear, and formally defined rules to follow for constructing a business model to eliminate ambiguity and confusion, however this methodology is neither graphical nor is the output easy to understand. The 22 results of an OBPM look like computer code. As noted, there are cognitive advantages to visually representing information. "The human ability to extract information from visual scenes is much more fundamental than is our ability to manipulate data verbally or arithmetically" (Zhang 1998, citing Schwartz and Howell, 1985) "Model diagrams enable the analyst to extract process, understand, and respond to much relevant information. The transfer of information is fast, accurate, and the user learning curve is minimized; the analyst can thus build a conceptual model of the problem with fewer perceptual errors" (Zhang 1998, citing Brown 1988) "The diagrams can also serve as the interface between a domain analyst with his/her customers. The visual model is indeed worth thousands of words in terms of communicating with customers." (Zhang, 1998, p.22) Another problem with the current incarnation of the OBPM algorithm is that it is based upon a set of modeling integrity rules. The premise behind basing the algorithm on a set of integrity rules is that if the algorithm is followed correctly then any models generated using the algorithm will be correct. The flaw with this approach is that not all of the modeling integrity rules are used in the OBPM algorithm. The rest of this chapter is organized as follows: to begin there will be a discussion of the OBPM algorithm, this will be followed by a discussion concerning object-orientation and its application to OBPM, a notation for creating object-oriented activity-based process models will be introduced, and then a process for creation and validation of object-oriented activity-based process models will be presented. 23 3.2.1 O B P M Algorithm The OBPM algorithm is presented here. It has three parts; the main routine, the affected thing subroutine, and the decompose subroutine. The algorithm serves to identify: Events (what triggers instability in things in a process), Activities (what happens in response to events), Operations (the transformations that occur), and Resources (what is used during transformations) The main routine begins the modeling process. The events that trigger the process from the environment are identified. For each triggering event the things in the environment that are affected should be identified. Then the subroutine affected thing should be invoked for each thing that was affected. The affected thing subroutine begins with if a thing is changed from a stable to a stable state it is considered to be a resource and this subroutine exits. If the thing changes from stable to unstable it is considered to be an agent. Agents invoke the decompose subroutine. When the decompose subroutine exits and returns to the affected thing subroutine, the affected thing subroutine will exit and return to the main routine. The decompose subroutine identifies the sequence of events that occur from when a thing is affected (becomes unstable) until the agent is finished changing (becomes stable). The entire sequence of events is known as an activity. Each event in the activity is an operation. If in the course of the activity other agents are affected then these agents will themselves invoke the affected thing subroutine. 24 The output of the algorithm is a series of lists. The first list is the agent list, those agents that participate in the process. For each agent there is also a list of the activities the agent performs and their sequence (called the Activity List) and a list of the operations for each activity and their sequence (called the Operations List). The algorithm also produces a resource list of the resources that are used in the process. The last list the algorithm produces is an event list which is a list of the events that happen in the process, and their sequence. 3.3 Object-Oriented Activity-Based Process Modeling Grammar The concept of object orientation has been applied to MIS and is the basis of object-oriented systems. There also exist graphical modeling languages that are object-oriented such as OOEM. Object-oriented modeling languages can also be used to create a model of what is occurring independent of any implementation. It has been noted that object-orientation allows a process model to be comprehensive, understandable, changeable, adaptable, and reusable (Hui, 1997).As shown in table 3-1, Wand and Woo (2002) previously mapped ontological constructs and premises into object constructs and premises. 25 Ontologically-Based Concept or Premise Object Construct Principles The world is made of things possessing properties* Objects and their properties are the fundamental modeling constructs* All things change and all changes are tied to things* Encapsulation: state and behaviour are combined* Things can combine to form composites* Objects can form composite objects* Things can affect each other's state evolution* Objects interact* Things can be categorized into classes defined by properties* A class is a set of objects sharing a group of the same properties* Concepts Thing* Object* Actor Object with services Non-actor Object with no services Actuator Object with at least one joint state variable that it owns, and one service to modify it. Propagator An actuator object with at least one joint state variable that it owns shared with an Actor Object Properties* Not Modeled directly. See attributes* Attribute functions* Attribute* Attribute representing inherent property* Internal state variable* Attribute representing mutual property* Joint state variable* State* State (attribute values)* Internal transformations* Services* Composite thing* Composite objects* Interaction Communication (via requests)* x acts-on y* ' • x and y have a shared state variable modifiable by x only* ..• , • Functional schema* Definition of a class* Event* State change* • • ' ' External event* Request* Internal event* Action (execution of a service)* Law • • • Service restriction Process System Table 3-1 Mapping Ontological Constructs And Premises To Object-Oriented Constructs And Premises *denotes original mapping (Wand and Woo 2002) As defined (Wart, Wand, & Woo, 1993) there are three general concepts of object orientation: classes and objects, association, and object communication. 3.3.1 Classes and Objects A class is a collection of things that share common features. An object is an instance of a class. An object encapsulates its attributes and behaviours, that is, they are included with the definition of the object. For example, my pet trout is an instance of the class trout. We 26 know that my pet trout has fins (property) and will swim (behaviour) from the definition of a trout When an object is defined only the properties and behaviours relevant for our purposes is defined. Returning to our example, if we are hungry fishermen our definition of a trout (probably) does not include any mention of the light refraction index of the skin of a trout, but rather the fact it is an edible fish. Thus a definition of an object or class is not always perfect. This is the object-oriented concept of abstraction. As mentioned above, objects have properties. These are referred to as attributes. Attributes only possessed by an object are internal attributes, attributes shared with another object are mutual attributes. When the values of attributes are measured the result is the state of the object. For example if our trout has fins, and the measure of the fins is broken, we can say the state of the trout is injured. Services are the behaviour of objects. Services are what change attribute values. They have a well defined interface that is used to change attributes. When our trout friend swims the value of the attribute stomach contents will decrease and our trout will become hungry. Encapsulation refers to storing the attributes and services (i.e. the state and behaviour), of an object together so that other objects do not need to worry about unnecessary information. This means that only the objects behaviour can access or change its state. 27 The only way one object can find information about another is to send a request to the object. If our trout wants to know if another (bigger) trout is hungry it can swim by its line of sight (sending a message, "Hey, I'm a smaller trout"). If the other trout tries to eat our trout in response, then our trout knows the bigger trout is hungry. Instantiation is a particular occurrence of an object that can be distinguished from other occurrences of an object. My pet trout is an instance that is different from all the other trout objects that may exist, since its attribute denoting ownership has a value of "me" Services and attributes are always inherited from the class to which an object is a member of. Hence all instances of trout will have the attributes and services of fish. However, different instances can have different values for their attributes. For example my pet trout may have speckled as its skin color, whereas your trout may have rainbow as its skin color. 3.3.2 Association Objects can associate, that is, two or more objects can have a relationship. The most important relationships between objects are aggregation and classification. An aggregation is a collection of component objects. The aggregation can be disaggregated into its component objects. Aggregations typically possess characteristics that are not present in the individual components themselves. A computer is an a^ggregation of a "CPU, monitor, keyboard, etc. Combined they possess a new characteristic, processing power, that is not present in the components. A computer can then be disaggregated (disassembled) into its components. Classification refers to being able to create generalizations and specializations to represent knowledge about classes. A generalization refers to identifying common properties of things to assist in the creation of abstractions. For example, if we notice my computer has a 17 inch monitor and your computer has a 15 inch monitor we may decide a computer class can be created with the attribute monitor size. A specialization is a more specific class (subclass) that inherits everything from its parent (superclass). Typically a subclass has attributes and services that the superclass does not. A laptop can be a subclass of the class computer. It has all the properties of a desktop, for example processing speed, and properties that a desktop does not have such as battery type. 3.3.3 Object Communication Objects interact to request other objects perform services that the requesting object cannot perform. The services can be used to enforce constraints on the relationship. A student can request a professor to open a classroom. A professor will not open the classroom if it does not know that the student is a member of the professor's institution. This enforces the constraint that professors only open classrooms to registered students. 29 Objects can interact with those both inside and outside the system. External objects make requests to internal objects. Internal objects can make requests to both internal and external objects. 3.3.4 The Combination of OBPM and Object-Orientation Object-orientation, obviously, centers around the object. OBPM centers around the agent. According to the BWWP ontology an agent is a thing that possesses properties and undergoes change via operations. An object possesses attributes and performs services. A service is how an object changes. The combined construct of an OBPM agent and an object will henceforth be referred to as an object-oriented activity-based process modeling agent (or just agent for short). The aforementioned links are still ambiguous and unclear. Figure 3-1 is an illustration of what we now call an agent. to I Change in other Agents ; : (Change Propagation) Figure 3-1 A B P M Agent Here is a brief definition of Figure 3-1. An agent "communicates" with another agent via changes. This "communication" can take on two forms. First an agent can have an interface attribute changed by another, agent. Second an operation of an agent may change an interface attribute, of another agent. A change to an interface attribute may trigger an operation. An operation's triggering or output may be governed by laws. An operation The Agent Internal Attributes Laws Change ] Operation Triggers Incoming Interface Attribute Change du another Ag Outgoing Interface Attribute 30 changes either an internal attribute or an outgoing interface attribute. Each of the constructs in figure 3-1 is discussed below. 3.3.4.1 Attributes Recall that in BWWP attributes are used to represent the properties of a thing, which in turn represent the state of the thing. In A B P M attributes can be used to represent the state of the Agent. In figure 3-1 there are two main kinds of attributes; interface attributes and internal attributes. Internal attributes are not known to other agents and can only be accessed or changed via the services of the agent. This demonstrates object-oriented principles of both encapsulation and object independence. Interface attributes can be accessed (changed) by other objects. Interface attributes model mutual properties of things. Agents interact with each other via changing the value of mutual properties. Therefore all interface attributes have two agents associated with them; an agent doing the changing (an outgoing interface attribute), and the agent being changed (an incoming interface attribute). Internal attributes are solely possessed by an Agent and are unknown to other agents. An internal attribute is what is changed by an internal event in an agent. 3.3.4.2 Operations and Change Propagation Changes result from interactions between agents. There are two types of changes possible; a change in an agent caused by another agent and a change in an agent caused by itself. For our purposes a change can be defined as altering the value o f an attribute. A n agent causes change in another agent by changing the value of an outgoing interface attribute it owns which is associated with the incoming interface attribute o f another agent. A n agent causes change in itself by altering the value o f an internal attribute. Accord ing to the object oriented literature a service is how an object does anything. Hence for an object to affect itself or another object it needs a service. That said, in O B P M agents affect themselves or each other via changes. Changes are carried out as operations. Hence a change is implemented through an operation in A B P M . When an agent has an interface attribute changed resulting in instability the agent performs one or more activities to become stable. A s part of an activity the agent may change one or more agents causing them to become unstable. These agents may then change others, and so on. This is known as change propagation. 3.3.4.3 Activities A n activity is what happens in an agent from the time when it becomes unstable to the time when it is stable. A l l agents are initially stable. Instability is caused by an incoming interface attribute being changed. When an activity occurs operations occur until the agent is stable. The operations w i l l change some combination of internal and outgoing interface attributes. Thus an activity is made o f operations that use or modify attributes. 32 3 . 3 . 4 . 4 Laws According to BWWP laws are properties of a thing. They restrict how a thing can change. In the case of A B P M the restriction is on what operations can occur in an agent. Recall from chapter 2 in the discussion on the B W W ontology, "The set of values for the attributes of a thing comprise the state of the thing. A conceivable state space for a thing is the set of all possible states a thing may ever assume. State laws serve to restrict the values of the properties of a thing to a subset of the conceivable state space. State laws must enforce a restriction due to either natural or human laws. A law is a property. For example, most bank accounts have the restriction (state law) that the balance must be greater than or equal to zero. This is due to the human law that people are only allowed to spend up to the total amount of money that is in their bank account. The lawful state space of a thing is the set of states that exist for a thing that comply with its state laws. A lawful state space is usually a subset of the conceivable state space. A transformation is a state (attribute) change from one state to another state. A lawful transformation defines the events that are lawful for a thing. The lawful event space is usually a subset of the event space, and defines those events in a thing that are lawful." For A B P M , a state law still enforces constraints. They dictate the constraints on what the output from an operation is, the output of an operation is an alteration of the value of a property. Thus they restrict the values a property can be altered to and by extension they restrict the lawful state space. 33 For A B P M transformation laws enforce constraints on which changesxould occur. They restrict the possible set of transformations (i.e. operations, since operations are how the transformations are carried out) to a set that are deemed lawful in an agent. In short, attributes and operations represent the properties and behaviour of agents, while the state and transformation laws are the constraints on properties (attribute values) and behaviour respectively. 3.3.5 Mapping Summary Table 3-2 summarizes how the O B P M constructs will be mapped to Object constructs to create Object-Oriented OBPM Constructs. 3.4 Meta-model of A B P M Figure 3-2 presents the A B P M meta-model. The metamodel can be used to show the relationships between the constructs, as discussed above, in a condensed manner. The meta-model shows theABPM constructs as rectangles. The relationships between constructs are shown using arrows. The cardinality numbers indicate the requirements on the relationship. The inverted *F show that a construct exists in both the generalized and specialized role. 34 OBPM Construct Object Construct ABPM Construct Agent Object Agent Actor Specialization of an Object Agent Resource Specialization of an Object Resource (An Agent with no services) Properties Attribute Attribute Intrinsic Property Internal Attribute Internal Attribute Mutual Property Shared Attribute Incoming Interface Attribute + Outgoing Interface Attribute Composite thing Composite object (aggregation) Composite Agent(aggregation) State Attribute values Attribute values Agent in the domain of interest Internal object Internal Agent Agent outside the domain of External object External Agent interest Operation Service Operation Activity The events that occur when an The events that occur when an agent Object becomes unstable becomes unstable Transformation law Service restriction Transformation law State law Service restriction State law Event A trigger of a service A trigger of an operation Internal event An event inside the object Change of an internal attribute, causing a transformation in the performed by an operation object External event An event outside the object Change of an incoming interface causing a transformation in the attribute object . X affects Y X changes the shared attribute of XandY X changes its outgoing interface attribute which is tied to the incoming interface attribute of Y Stability No services required by an object No operations required by an Agent Inherited Attributes Inheritance Super and Sub agents Process Propagation Process Table 3-2 Mapping Summary The meta-model is explained as follows: an agent is either internal or external. Internal agents perform activities. An internal agent performs one or more activities, but the activity is performed by only one agent. Activities consist of operations and attributes. An activity must consist of one or more operations; however the operations occur in only one activity. An activity involves changes to two or more attributes; however the attributes are affected in only one activity. 35 An operation changes one or more attributes, the attributes can only be changed by one operation. Figure 3-2 ABPM Meta-Model An activity may be governed by zero or more laws, the laws govern only one activity. A law is either a state law or a transformation law. State laws,govern attributes, one state law governs one attribute. A transformation law governs one or more operations, the operations are governed by only one transformation law. Attributes are either internal attributes or interface attributes. Interface attributes can further be broken up into incoming interface attributes and outgoing interface attributes. 36 Internal agents have two or more attributes, the attributes are only possessed by one agent. Internal agents have one or more incoming interface attributes, the incoming interface attribute(s) is(are) only possessed by one agent. Internal agents are required to have at least one incoming interface attribute since something external to the agent is what initiates change in the agent. The requirement of two or more attributes refers to an internal agent needing to both be changed (i.e. have one incoming interface attribute), and then change something else (in itself or another agent). Hence at least two attributes, one incoming interface + at least one other attribute it changes in response to being changed. If an internal agent were to only have incoming interface attributes then it would be a resource. An external agent has one or more interface attributes that are changed, the interface attributes are possessed by only one agent. Resource are changed by agents. An agent can change one or more resources, but the changes are performed by only one agent. A resource only has incoming interface attributes. A resource can have one or more incoming interface attributes, but the incoming interface attributes are possessed by only one resource. 3.5 Graphical Representation We are now ready to introduce our graphical constructs for A B P M . The following subsections will first show the construct then follow with an explanation of the construct. The graphical constructs can be used to assist the user in following the A B P M modeling process (as presented in section 3.6 below). 37 3.5.1 Domain Representation Agent 1 Agent 2 Agent 3 Agent 4 Figure 3-3 Domain Representation If we are trying to show the agents that interact in a process, we need to consider the domain they will interacting in, i.e. the scope of the process. The domain in which they interact will be represented using swimlanes, with the swimlanes themselves being the boundary of an agent. 3.5.2 External Agent External Agent Agent 1 Agent 2 Agent 3 Figure 3-4 External Agent Representation The name of an external agent is doublelined. External agents are what cause the initial event(s) that occur.outside the domain of a process. They perform activities that change (affect) a property of a thing in the domain. An external agent may also have one or more attribute changed by.an agent in the system. The only information we need to represent in 38 the swimlane of an external agent are the incoming interface attributes from the system and the outgoing interface attributes to the system. Incoming interface attributes follow the notation: agent that changed the attribute:attribute changed Incoming interface attributes are how an agent receives a change from another agent. Outgoing interface attributes follow the notation: attribute changed::agent changed Outgoing interface attributes are how an agent initiates change in another agent. 3.5.3 Internal Agent External Agent Internal Agent 1 Internal Agent 2 Activity 1 Name Affected Attributes Stale Law(s) Incoming Interface Attribute(s) Internal Attribute(s) Outgoing Interface Attributes(s) Activity 1 Name Affected Attributes . State Law(s) Incoming Interface Attribute(s) Internal Attribute(s) Outgoing Interface Attributes(s) Activity 1 Operations Transformation Laws Operation Operation ' Activity 1 Operations Transformation Laws Operation Operation Activity n Name Affected Attributes Activity n Operations Activity n Name Affected Attributes Activity n Operations Figure 3-5 Internal Agent Representation Since external agents outside the domain affect agents in the domain we need internal agents that, represent the agents within the domain. The name of an internal agent is singly-lined. 39 A n activity consists o f attributes, operations and laws. In particular, an activity is the attributes that are changed and the operations that change them in response to the agent receiving a change from another agent with the laws that dictate contraints on how they are changed. A n agent can receive an infinite number of changes from other agents hence there could be an infinite number of activities. In keeping with the concept of encapsulation in the representation o f an activity we keep the attributes that are changed and the operations that change or use them together. Hence we outline each activity to illustrate the encapsulation. The first information included in the affected attributes o f an activity is the state laws on the attributes. A state law dictates i f there are any restrictions on the values an attribute can take. This is a freeform box of text before the incoming interface attributes o f an activity. Incoming interface attributes fol low the notation: agent that changed the attribute: attribute changed Incoming interface attributes are how an agent receives a change from another agent. Internal attributes are changed solely by the internal agent. They are changed v ia operations. 40 Outgoing interface attributes follow the notation: .. _. attribute changed::agent changed Outgoing interface attributes are how an agent initiates change in another agent. They are changed via operations, which will also change the associated incoming interface attribute of another agent. When we look at the operations for an activity the first thing that is represented is the transformation laws. Transformation laws dictate which behaviours (operations) occur under what conditions. A transformation law is represented by a freeform box of text before the operations of an activity. An operation is what changes an attribute. An operation may change several attributes but must at least change one attribute. Hence for every attribute that changes there must be an operation that changes it. The attributes that an agent changes are its internal attributes and its outgoing interface attributes. Thus for every internal attribute and every outgoing interface attribute there is an operation that changes it. The temporal sequence of attributes and operations should be kept the same. If the laws and incoming interface attributes are removed from an activity, an attribute in position X of the affected attributes list, should have the operation that changes it in position X of the operations list. This is in keeping with the general notion of a process. 41 We also need to consider how do we number activities? Our solution to this is that we number them according to how many activities there are in an agent. So if an agent performs nine activities, they are numbered one through nine. Thus different agents will have their own activity one. Typically process models will give each'activity its own number thus there will only be one activity number one, etc. 3.5.4 Resources External Agent Internal Agent Resource Incoming Interface Attribute Incoming Interface Attribute Figure 3-6 Resource Representation Resources are those things that only are changed by other things. A resource is indicated by a treble line around the name. When they are changed they go from one stable state to another stable state. The have no internal transformations. Hence they have no operations. Thus a resource will only have attributes of the notation: agent that changed the attribute: attribute changed 3.5.5 Agents Sharing A Mutual Attribute The arrow in figure 3-7 designates that the agents share an interface attribute. The agent at the tail of the arrow changes the outgoing interface attribute it has while the agent at the head of the arrow is the receiver of the change via its incoming interface attribute. A block arrow designates that an external agent changes some thing in the system. External Agent Outgoing Interface Attribute Internal Agent 1 Activity 1 Name Affected Attributes State Law(s) Incoming Interface Attribute Internal Attribute Outgoing Interface Attribute Outgoing Interface Attribute [Activity 1 Operations Transformation Laws lOperation (operation •— ^Operation Activity n Name Affected Attributes Incoming Interface Attribute [Activity n Operations Internal Agent 2 Activity 1 Name Affected Attributes State Law(s) Incoming Interface Attribute internal Attribute (internal Attribute [Activity 1 Operations Transformation Laws [Operation nation Figure 3-7 Agents Sharing A Mutual Attribute Representation Recall, we are creating models at the class level. When an arrow exists between two activities in the same agent it shows one instance of agent can change another instance of the same agent. In the event an agent can change the exact same incoming interface attribute of another agent from different activities, there are be multiple outgoing interface attributes but only one incoming interface attribute that is changed. In this case,the arrows are merged into ! one doubled arrow as illustrated in figure 3-8. " i Internal Agent 1 Activity 1 Name Affected Attributes Outgoing Interface Attribute Activity 1 Operations . Operation | Activity n Name Affected Attributes Outgoing Interface Attribute Activity n Operations Operation ' - — Internal Agent 2 Activity 1 Name Affected Attributes State Law(s) Incoming Interface Attribute Internal Attribute Internal Attribute [Activity 1 Operations Transformation Laws lOperation [operation Figure 3-8 Different Operations changing one incoming interface attribute 4 3 3.5.6 Composite agents External Agent Internal Agent Activity 1 Name Affected Attributes State Law(s) Incoming Interface Attribute^ Internal Attribute(s) [Outgoing Interface Attributes(s)! |Activity 1 Operations Transformation Laws Operation Operation lActivity n Name lAffected Attributes |Activity n Operations Figure 3-9 Composite Agent Representation Composite Agent Activity 1 Name Affected Attributes State Law(s) Incoming Interface Attribute^ Internal Attribute(s) [Outgoing Interface Attributes(s)! (Activity 1 Operations Transformation Laws Operation Operation lActivity n Name lAffected Attributes lActivity n Operations Composite agents in their composite v iew appear the same as any other agent except the lines around their name are,dashed. If an activity in a composite agent is underlined the activity is an emergent activity that would not exist without the aggregation o f components. Act iv i t ies that are not underlined are those'from component agents. The composite agent v iew is used since we may not be interested in all the details o f a composition. In those situations where we need to know the details o f a composite agent a composite agent can be decomposed down into its component agents: When a composite agent is created from component agents it must possess emergent attributes, i.e.its own attributes that are not part o f any component agent, thus they need 44 their own operations as well. The emergent attributes and operations would not exist unless the agents are aggregated. Figure 3-10, illustrates how a composite agent can be added to a diagram without having to rearrange the entire diagram. The composite agent lists its component agents, while the components list that they are part of a composition. Agent 1 Composite Agent Agent 2 Component Agent 1 Agent 3 Component Agent 2 Component Agent 1 Component Agent n Composite Agent i Composite Agent • Activity 1 Name Affected Attributes State Law(s) Incomino. Interface Attribute(s) Internal Attnbute(s) Activity 1 Name Affected Attributes Sfafe Law(s) Incoming Interface Attribute(s). Internal Attribute(s) Outgoing Interface Attributes(s) Activity 1 Name Affected Attributes State Law(s) Incoming Interface Attribute(s) Internal Attribute(s) Outgoing Interface Attributes(s) Outaoina Interface Attributes(s) Activity 1 Operations Transformation Laws Activity 1 Operations Transformation Laws Operation Activity 1 Operations Transformation Laws Operation Operation Operation Operation Operation Figure 3-10 Adding A Composite Agent The decomposition of a composite agent allows us to see how the agents that make up the composite agent interact, and what attributes and operations emerge from the composition in one figure. Figure 3-11 is an example of the decomposition for a composite agent. Ideally we would include a diagram like figure 3-11 and not figure 3-10 when interested in the composition of a composite agent. HoweverKvhen it is not feasible • i ! i to rearrange the diagram to have all the component agents side by side the arrangement in figure 3-10 can be used; : The decomposition as shown in figure 3-11 can also be used to verify the integrity of a composite agent. By only including the composite and its components we can verify that 45 all incoming changes and outgoing changes are received and generated (respectively) by the composition, or its components. Composite Agent Component Agent 1 Component Agent 2 [Activity 1 Name [Affected Attributes \State Law(s) Incoming Interface Attribute(s) Internal Attribute(s) [Outgoing Interface Attributes(s)| [Activity 1 Operations Transformation Laws [Operation j [operation Activity 1 Name Affected Attributes State Law(s) Incoming Interface Attribute(s) Internal Attribute(s) Outgoing Interface Attributes(s)| [Activity 1 Operations Transformation Laws [Operation [operation Activity 1 Name Affected Attributes -State Law(s) ' Incoming Interface Attributefs) Internal Attributed) Outgoing Interface Attributes(s) Activity 1 Operations Transformation Laws Operation Operation Figure 3-11 Decomposition Of A Composite Agent When decomposing a composite agent into its components the following must specified: • Which incoming changes go to which component agent? • Which incoming changes go to the composite agent (are emergent)? • -Which outgoing changes are generated by which component agent? • Which outgoing changes are generated by the composite agent (are emergent)? • Which changes do the components use to interact with each other? These changes are not present in the composite .view of the agent. • Which internal attributes belong to the composite and the components? < • What laws exist on the incoming and outgoing changes of the components? When the attributes are reassigned to either the composite or component the operation that is associated with that attribute is reassigned as well. Component agents may require new attributes and services not present in the composite agent to model the interaction of the component agents. 46 3.5.7 Superagents And Subagents As shown in figure 3-9, the agent on top is the superagent(i.e. the generalization), the agents on bottom are subagents(i.e. the specialization). Attributes and operations in the superagent are inherited by the subagent. Inherited attributes and operations are not shown in the subagent. Subagents have attributes and operations possessed only by that subagent. A subagent's inherited attributes and operations can never be different from the attributes and operations of the superagent (from the definition of a subclass), hence we do not need to show them. Figure 3-13 shows how to represent superagents and subagents when it is not feasible to rearrange the diagram to have them side by side. Super Agent Activity 1 Name Affected Attributes . State Law(s) Incoming Interface Attribute(s) Internal Attribute(s) Outgoing Interface Attributes(s) Activity 1 Operations Transformation Laws • • Operation -Operation Sub Agent 1 Activity 1 Name Affected Attributes Sfafe Law(s) Incoming Interface Attribute(s) Internal Attribute(s) Outgoing Interface Attributes(s)| Activity 1 Operations Transformation Laws '• [Operation [Operation [Activity n Name lAffected Attributes [Activity n Operations Sub Agent 2 [Activity 1 Name lAffected Attributes State Law(s) Incoming Interface Attribute(s) Internal Attribute(s) lOutgoing Interface Attributes(s)! |Activity 1 Operations Transformation Laws lOperation [Operation Kctivity n Name ffected Attributes [Activity n Operations Figure 3-12 Superagent and Subagent Representation 47 Agent 1 Superagent : Agent 2 lActivity 1 Name [Affected Attributes Srafe Lew(s) Incoming Interface Attribute(s) nternal Attribute(s) lOutgoing Interface Attributes(s; |Activity 1 Operations Transformation Laws lOperation loperation SubAgent 1 Superagent lActivity 1 Name [Affected Attributes State Law(s) Incoming Interface Attribute(s) Internal Attribute(s) lOutgoing Interface Attributes^ ^Activity 1 Operations Transformation Laws lOperation [Operation Agent 3 Subagent 2 Superagent lActivity 1 Name . lAffected Attributes , ]S!a(e Law(s) ncoming Interface Attribute(s) llnternal Attribute(s) lOutgoing Interface Attributes^; (Activity 1 Operations Transformation Laws lOperation loperation Figure3-13 Creating A Superagent 3.5.8 An Alternate Notation There are two situations where it is beneficial to have an ABPM with fewer details. 1. The details are not known 2. The diagrams are large and unwieldy We can handle both situations using Agent Templates (AT). An A T table can be used to store some or even aU the pertinent details about the agent and an agent with less or even no details appears in the A B P M . Since we are focusing on change propagation the two suggested compressions are either retaining only the interface attributes that are changed in an activity (as per figure 3-14), or compressing an activity totally down to its name(as per figure 3-15). External Agent Internal Agent 1 Outgoing Interface Attribute f_ Incoming Interface Attribute * lActivity 1 Name lAffected Attributes ncoming Interface Attribute lOutgoing Interface Attribute — | Internal Agent 2 lActivity 1 Name lAffected Attributes ncoming Interface Attribute {Outgoing Interface Attribute Figure 3-14 Example Compressing Down To Agent Interactions External Agent Internal Agent 1 Internal Agent 2 i Outgoing Interface Attribute | jj/JActivity 1 NameT- lActivity 1 Name | Incoming Interface Attribute « i -Figure 3-15 Example Compressing Down To Activity Name 48 The internal agent template is presented in Figure 3-16. It w i l l be f i l led in using the notation developed above.. _ ... Agent Attributes Operations Activity State Law Interface Attributes Internal Attributes Transformation Law Operation 1 2 n Figure 3-16 Internal Agent Template The Resource template is presented in figure 3-17. Since it only has incoming interface attributes all entries w i l l be in the form: agent that changed the attribute: attribute changed . . . . Resource .: Incoming Changes Figure 3-17 Resource Template With external agents less detail is needed than with internal agents. The only information of interest is the incoming and outgoing interface attributes that are altered. The external agent template is presented in figure 3-18. Incoming interface attributes w i l l fo l low the form: agent that changed the attribute:attribute changed Outgoing interface attributes w i l l fo l low the form: attribute changed: :agent being changed 49 Agent Incoming Changes Outgoing Changes Figure 3-18 External Agent Template In a composite agent the agent template has extra rows. The extra rows indicate which agents are components o f the composite agent. The component agents are the same as any internal agent and thus have the same agent template as an internal agent. The composite agent template has two views (using the same template). The first view is just of the activities of the composite agent (i.e. the emergent attributes and operations). The second view is the 'composited' view, that is, all the activities o f the components that are included in the composite and its emergent activities. The emergent attributes and operations would be italicized in the composite view to distinguish them from component attributes and services ( i f the agent has been decomposed). Figure 3-19 shows the composite agent template. Agent Component Agent 1 Component Agent 2 Component Agent n Attributes Operations Activity State Law Interface Attributes Internal Attributes Transformation Law Operation 1 2 n Figure 3-19 Composite Agent Template The agent template for superagents is not different from the internal agent template. The agent template for subagents has an extra row to indicate the superagent from which it is 50 derived. The attributes and operations that are shown are those that are unique to the subagent. The subagent template is shown in figure 3-20. Agent Superagent: Attributes Operations Activity State Law Interface Attributes Internal Attributes Transformation Law Operation 1 2 n Figure 3-20 Subagent Agent Activity Template 3 . 5 . 9 Allowed Interactions This section summarizes how the constructs are allowed to interact. External agents: Initiate change in an agent in the system. Receive a change from an agent in the system. Refer to figure 3-21 E x t e r n a l A g e n t A g e n t 1 A g e n t 2 E x t e r n a l A g e n t A g e n t 1 A g e n t 2 - • Initiating a change in the system Receiving a change from an agent in the system Figure 3-21 Allowed Interactions For An External Agent Resources: Receive a change from an agent in the system. Refer, to figure 3-22 R e s o u r c e A g e n t 1 A g e n t 2 Receiving a change from an agent in the system Figure 3-22 Allowed Interactions For A Resource Internal agent: Initiate change in both internal and external agents. Receive a change from internal and external agents. Change a resource. Refer to figure 3-23 51 External Agent Agent 1 Agent 2 External Agent Agent 1' Agent 2 Changing an agent outside the system Changing an agent in the system External Agent Agent 1 Agent 2 External Agent Agent 1 Agent 2 Receiving a change from outside the system Receiving a change from an agent in the system Resource Agent 1 Agent 2 Changing a resource Figure 3-22 Allowed Interactions For An Internal Agent 3.6 A B P M Modeling Process The main concept to keep in mind when generating an A B P M is change propagation. That is, when one agent is changed, this change will cause the agent to change other agents, which will lead to other agents being changed, and so on. Eventually, there are no more agents that are changed and the process ends when all the agents that are changed become stable (stop changing). Frequently process-modeling languages lack a well-defined method for generating a process model (Huckvale and'Ould,1994, Wand and Woo, 1999,'Wang 2002). One of the goals behind the development of OBPM was to eliminate problems with existing process modeling languages. Hence we must provide a clear and unambiguous method to 52 create an A B P M . Thus we will present an algorithm for the graphic representation of ABPMs. This algorithm.is based upon a set of modeling rules. 3.6.1 Modeling rules At this point we need to introduce a set of rules (Wand and Woo, 1999) and assumptions to address the following points: 1. What is the scope of the model? 2. What agents should be included in the model? 3. What resources should be included in the model? 4. What agent properties and operations should be included in the model? 5. When to include composites agents? 6. When to subclassify agents? 7. When to begin a new activity? Due to the fact that they are developed specifically for object-oriented models that are ontologically based (albeit for enterprise modeling) we will use and adapt the rules developed elsewhere (Wand and Woo 1999) to answer the questions. Only when the rule is changed from its original intent will we define its development. Rule #1: The scope identification rule This rule is used to define what should be included in the process model of the system. It is based uponthe belief that anything that happens in the system is in response to something happening outside the system. That is, something outside the system affects something in the system causing the system to become unstable. This is the only way a system can become unstable. Once a modeler decides what events happen which are 53 external to the system, all direct and indirect actions due to the external event are in the scope of the system. Thus the rule reads as follows: Rule 1 (scope): The aspects of the system to be modeled are all and only those needed to represent the effects of the relevant external events5. Rule #2: The affected thing identification rule An event is a change. From our ontological foundation only things change. To be included a thing has to either be changing another thing or changed by another thing. Thus the rule reads as follows: Rule 2a (agent identification): The agents included should be those that are either generating changes in the system, or are responding directly and indirectly to the external changes to the system. Rule 2b (Resource identification): The resources included should be those that are changed by agents in the system. The agent identification rule highlights two kinds of agent. External agents are outside the system that either change something in the system or are changed by something in the system. Internal agents are the agents that make up the system. An internal agent must be changed by at least one other agent. The resource identification rule indicates that only the resources used by internal agents should be included Rule #3: The operation inclusion rule Based on our mapping operations represent transformations that happen to an agent. Operations will be invoked by an unstable agent to change an attribute in an attempt to become stable during an activity. An agent becomes unstable when the prerequisite interface attributes (i.e. incoming interface attributes) of an activity have been changed. Rule 3a (operation inclusion): An operation will be included in an activity if it is invoked as a result of an agent attempting to become stable. Rule 3b (transformation law inclusion): A transformation law will be included if it affects what operations occur. Rule #4: The attribute inclusion rule Since all activity in a system is initiated due to an external change, only those attributes that are part of the activities due to the external change should be included. The attributes that are part of an activity are those that either initiate the activity or are changed by operations during an activity. The attributes must be used or modified by an operation. Rule 4a (interface attribute inclusion): an interface attribute will be included if it is known by and shared between two things. An outgoing interface attribute can only be possessed by an agent, whereas an incoming interface attribute can be possessed by an agent or a resource. Rule 4b (Internal attribute inclusion): an internal attribute will be included only for those operations that do not act upon an interface attribute as defined from 4a. An internal attribute is internal to an agent. Rule 4c (state law inclusion): a state law will only be included if it restricts the values an attribute can be changed to. 5 This is the original Wand and Woo (1999) .definition of the rule. 55 Rule #5: The attribute ownership rule Since properties belong to. things, every attribute is owned by only one agent. . Rule 5 (attribute ownership): For every attribute in the model there is exactly one agent that can modify it. For an internal attribute, the agent is the only one that can access the value of the attribute. For interface attributes the agent that possesses the outgoing interface attribute is the agent that modifies both the outgoing interface attribute and the incoming interface attribute (even though the incoming interface attribute is owned by another agent). Rule #6: The composite agent rule In some cases agents may need to be functioning together to respond to the changes resulting from the external stimulus since neither may be able to respond on its own. This creates emergent behaviour not present in either agent. Recall from ontology, that composite things have emergent behaviour. Rule 6 (Composite agents): A composite agent may be created only if it possesses emergent attributes not present in any of its components. A composite agent possesses all the attributes and operations of its components. Rule #7: The sub-classification rule In some cases agents may have properties that are very similar. It may be beneficial to create a superclass to simplify a model. Rule 7 (Sub-classification): A sub-class should be created only when it has properties not present in the superclass. A sub-class inherits all properties of the 56 superclass. In order to form a superclass two or more agents need to have some properties in common. Rule# 8: The new activity rule The first change in an agent is always the beginning of a new activity. An activity is a sequence of events and changes. Typically an actor becomes unstable, undergoes transformations, then becomes stable. Later on, other changes may cause the agent to yet again become unstable and the process repeats itself. Resources never become unstable, they are stable, changed by another agent, then are stable. Rule 8 (New activity) When an agent becomes unstable after being stable it is the beginning of a new activity. When an activity ends an agent is in a stable state. 3.6.2 Modeling Process If we follow the rules in a systematic manner we can produce an object-oriented activity-based process model. Further, there will be no need to check the integrity of the model since by following the algorithm correctly we ensure the model is correct. The guidance in applying the rules comes from the original O B P M algorithm. That is, we identify the changes generated external to the system, identify what is affected by the changes, and then analyze what has been affected. We can use this to follow changes as they propagate through the system. The following algorithm for creating ABPMs is designed to center around change propagation. 57 1. Identify the external agents. 2. For each external agent identify the changes generated. 3. For each change:. - — 3.1. Identify the agent or resource that was changed 3.2. If a resource was changed identify the incoming interface attribute 3.3. If the agent is an internal agent: 3.3.1 If this is the first change to an agent, or the last activity of an agent has gone through a sequence of instability-change-stability create a new " activity 3.3.2 Identify the incoming interface attributes that were modified 3.3.3 Identify any state laws that may restrict change 3.3.4 Identify any transformation laws that may exist for the incoming interface attributes 3.3.5 If an agent becomes unstable: 3.3.5.1 Identify the operations that may occur 3.3.5.2 Identify any transformation laws that may affect what operations occur 3.3.5.3 Identify the internal attributes that will be affected 3.3.5.4 Identify the outgoing interface attributes that were modified 3.3.5.5 Repeat steps 3.3.5.1 to 3.3.5.4 until the agent becomes stable 3.3.6 Repeat step 3 for each outgoing interface attribute of an agent that was changed in step 3.3.5.4 4 If needed identify super and subagents using the internal agents. 5 If needed identify composite and component agents using the internal agents. Table 3-3 shows the relationship between the rules, the A B P M algorithm, and the original O B P M algorithm. From this table we can see how .the proposed A B P M algorithm encompasses and expands upon the original O B P M algorithm. We can also see • that by following the A B P M algorithm we force the modeling rules developed above in section 3.6.1 to be followed 58 Step Rule Purpose O B P M algorithm I. Identify the external agents 2a Agent Identification 2. For each external agent identify the changes generated. 1 Scope Identification Main 3. For each change: 3.1 Identify the agent or resource that was changed 2a and 2b Agent and Resource Identification Main 3.2 If a resource was changed identify the incoming interface attribute 4a Interface Attribute Identification 3.3 If the agent is an internal agent: 3.3.1 If this is the first change to an agent, or the last activity of an agent has gone through a sequence of instability-change-stability create a new activity 8 Activity Identification Affected thing 3.3.2 Identify the incoming interface attributes that were modified 4a Interface Attribute Inclusion 3.3.3 Identify any state laws that may restrict change 4c State Law Inclusion 3.3.4 Identify any transformation laws that may exist for the incoming interface attributes 3b Transformation Law Inclusion 3.3.5 If an agent becomes unstable: 3.3.5.1 Identify the operations may occur 3a Operation Inclusion Decompose 3.3.5.2 Identify any transformation laws that may affect what operations occur 3b Transformation Law Inclusion 3.3.5.3 Identify the internal attributes that will be affected 4b Internal Attribute Inclusion • i . . ; , 3.3.5.4 Identify the outgoing interface attributes that were modified 4a Interface Attribute Inclusion' Affected thing 3.3.5.5 Repeat steps 3.3.5.1 to 3.3.5.4 until the agent becomes stable 3 Operation Inclusion Decompose 3.3.6 Repeat step 3 for each outgoing interface attribute of an agent that was changed in step 3.3.5.4 5 Attribute Ownership Decompose 4. If needed identify super and subagents using the internal agents. 7 Superagents and Subagents 5. If needed identify composite and component agents using the internal agents. 6 Composite and Component Agents Table 3-3 Relating The Rules To The ABPM And OBPM Algorithms 3.6.3 Model Integrity Once a model has been constructed it may be necessary to check the model if the model is semantically correct. Although the above modeling algorithm is supposed to ensure semantic correctness there may be other ways to generate an A B P M (such as from an O O E M , see the next chapter) that do not necessarily guarantee the model will be 59 semantically correct. We can check using a set of model integrity rules. The model integrity rules reflect the modeling rules (Wand and Woo, 1999). 1. Every change in the system should be able to be traced back to an initial change from an external agent. 2. Every agent must have at least one activity. 3 Every Activity must have at least one operation 4. Every resource must only have incoming interface attributes. 5. Every attribute is changed by only one operation. 6. Every activity can only have one incoming interface attribute unless governed by a transformation law 7. Every outgoing interface attribute must have a corresponding interface attribute and vice versa. 8. Every composite agent must possess emergent attributes and operations not present in the component agents 9. Every subagent must possess attributes and operations that are unique to the subagent and are not inherited from the superagent. 3.7 A n Example We recognize that this chapter has presented a fair bit of new ideas and concepts. To illustrate them, we will use the following example The ACME Warehouse Management Inc. Case6 ACME Warehouse Management Inc. offers storage facilities and redistribution services (between their different warehouses) across the nation. A customer can request space in a particular warehouse, request items to be transferred to another warehouse, or request withdrawal of items from a particular warehouse (even for items not stored there). For the purpose of this case, we only look at the activities involved'in processing cr withdrawal request. A customer contacts ACME headquarters to request a withdrawal. An office clerk checks whether the customer has the authority to withdraw the items. The clerk then passes the withdrawal request to the warehouse where the customer wants to pickup the items. If the warehouse does not have the items or does not have enough quantity of the items, the warehouse manager will contact other warehouses for the requested items. If the items are located the warehouse manager will ask the planner to arrange for transportation for the requested items. 6 Based on a case in I. Jacobson, ObjectrOriented Software Engineering, Addison-Wesley, 1992 The planner's responsibility is to schedule the company's truck fleet to accommodate requests for transportation, taking into account the existing schedule of each truck and its capacity. The warehouse manager will be notified whether the. transportation request can or cannot be satisfied. The warehouse manager will notify the office clerk if the request can be fulfilled or not, and the reason. The office clerk will notify the customer as to the status of the request (approved, or declined due to lack of authority, no inventory, or no transportation). The planner issues transport orders to truck drivers. After receiving a transport order, the truck driver informs the warehouse about the pickup of the items. The warehouse manager will make arrangements to have the items ready when the truck arrives. When the truck arrives at the warehouse the items are loaded. The truck driver then informs the next warehouse about the delivery. When the truck has arrived at the next warehouse, the items are unloaded. A warehouse worker finds space for the items and arranges to have them moved to the allocated space. The worker updates the warehouse's inventory information. Truck drivers are required to report the status of the truck and the delivery to the planner after each step. The customer will come to the warehouse on the required date to pick up the items. A warehouse employee will check all the necessary documents and will deliver the items with an-accompanying documentation to the customer. Supplemental description Once the office clerk has recorded the items to be withdrawn, he or she forwards the request to the manager (foreman) of the warehouse. The warehouse manager is responsible for directing the redistribution of items between warehouses: If the items are not all available in the warehouse, transport requests are issued: The warehouse manager fills out a redistribution form with the following information: items to be moved, place from which to take the items, warehouse to transport the items to, quantity to be moved, and'the date by when the redistribution must be done. The warehouse manager forwards the form to the planner to organize the interwarehouse transportation of the items. The items to be moved are marked as move-pending, and the planner initiates a plan to have the items at the appropriate warehouse at the given date. Once interwarehouse transport plans are finalized, transport requests are issued to the truck drivers. ••• • • • The truck driver alerts the warehouse manager of the time he or she will be at the warehouse to pick up the items. The warehouse manager gives appropriate requests to the warehouse worker on the date of delivery to have the items ready for when the truck is expected.-When the warehouse worker gets a request to fetch items, he or she, at the appropriate time, orders forklift operators to move the items to the loading platform. The forklift operators execute the internal warehouse operation. When the truck driver arrives, the driver notifies the warehouse worker to have the items loaded into the truck. The truck driver notifies the next warehouse manager when it is expected to arrive at the 61 next warehouse. The number of items in the current warehouse decreases, and the transport request is marked as on transport. When the truck has arrived at the next warehouse, the truck driver notifies the warehouse worker to unload the items. The truck driver signs off the job. The warehouse workers receive the items and determine a place for them in the warehouse. Forklift operators are told to move the items to the new place in the warehouse. When the truck driver confirms the delivery of the items, the records are updated to reflect the new place for the items. The transportation time is recorded and stored. The redistribution and interwarehouse transport request are marked as performed. The warehouse worker fills in an inventory update form and sends it to the warehouse manager for confirmation and update of the inventory database. When the customer has fetched the items the warehouse workers mark the withdrawal as ready. The items are removed (decreased) from the information system. Based on the case we develop the Compressed A B P M diagram showing only the agent interactions in figure 3-24. Figure 3-25 is the full uncompressed A B P M . I Mice C'lcrk::Ordcr Slulus ^ Office Cl«rk Adivii) 1 Withdrawal Request Affected Altrihultt 'CuMomcr::Withdraw ul request Order Status: :Customcr Activity 2 Notify about order A Heeled Attributes Warehouse-Order Status • Order Status: :C ustiwncr is and Docu mentation Warehouse Aclivilv I find Hem-, Affected Attributes •Office Clerk:: Withdrawal Rcuues Order Status: Ollice Clerk -hem Insistence:: Warehouse Activity. 2 Seaah>j» Hems Affected Ailributc* ..Warehouse Manager:;Item IMstcnec Search Result*:; W archouse Activilv 3 Decide if order can proceed A Heeled Attributes •Warehouse:: Search results Transport r'orm::l'lunner Order Status: :oltice clerk Activity 4 Notify about order status A Heeled Attributes I'lunncniTrunsnorl Schedule*—• (trder Slalus::nltice clerk Activity 5 Prepare ti* pickup Alleeted Attributes Truck Driver:: Pickup Not i Ileal i Activity 6 Load Truck Alleeted Attributes Truck Driver::Kcady lo L o a d * -On Trunspnrt::Truek Driver Activitj 7 Read) Receiving Alleeted Attributes Truck l>ri\er:!>ri>rH>ll'Nmilicatiorf-Activily H I Inliud truck Alleeted Attributes Truck Driver::Heady To United " Activity VFufil Pickup Alleeted Attributes <'ust omen: Arrival Items And I)oeurncntaiion::Cusiorocr Activity I Arrange Transport Affected Attributes • i . . Warehouse::Transport form Tninvporl Schedule:;Warehouse Transport orders::! ruck Driver — Activity 2 Record lime Alleeted Attributes Truck Driver:: I ruck Status Activity 1 Proceed to pickup Alleeted Attributes Planner:: Iran sport orders Pickup Notification::* archouse Truck Slaius^Plunncr Read) lo load:: Ware house 1 ruck Suius:: Planner >, ' Aclivilv 2 Proceed lo dropoff Alleeted Altrihules Waivhuuse: :On Transport Dn>potVNolilieation::WarehiHis Truck Status: :Planncr Ready lo Unload:Warehouse Truck Status: :Planner Figure 3-24 Compressed ABPM showing Agent Interactions 62 4Aciivii) ) Withdrawal Request AHecied Ariribiiics Customer: :Wi!hdraual request Authorization Status Withdrawul Request: Warehouse Order S |-""*:: ( '" '- '"" k ' f Office Clerk::Order Slums Acliv ii> I Operations Contact Warehouse und Noli I) Status arc mutual I) exclusive Auih.wit> Check Con tut I Warehouse : status Aciivii) 2 Nmil) ahoul rder Status Aci ivi i j 2 Operai Notify about slati A L ti\il> 1 l ind hen A Heeled Aiirihuk-s •Oltice l'lerk::Wiihdruual Request Inveniorv lnti>rmui i|'()^r1.li.t Coniuel I Utter Wan. house and Notil'v About Status are in liuillv exclusive Check Inventor) •Not if) about status Conlacl 1 Mlier Ware hot.SCS Aetiviij 2 !>vivch l-orlieim Alleeled Attribute* Warehouse;: Item l:\isierice In senl W) Information Search Resu lis: :Waa-house Activit) 2Ojvroiions Cheek Inventor) Not it) about search result* , | Aetivit) 3 Decide iforder can proceed | Alleetcd Attributes Warcht>use::Search result I transport Form:: Planner Or.1 T Sums••iillkc clerk | Actisitj } Operations ilact planner und rvHih ahoui I siaius are mutuall) exclusive Coniuel planner f Nolifv ahiul status Aciivii) 4 Nolif) about order status A fleeted Attributes Planner:: I ransporl Schedule * Order Staius::o11ke clerk Aclivit) 4 Operai f Nolil'v ahoul st ivitv 5 Prepare liir pickup ,-eted Attributes I I nick Driver.:l'ickup Noii lkuli J Iru Affect :led Attributes | Iruck Driver:;Read) lol.< torj In Tor mui it ui On Iraiispori"'!ruek Drive Aetivit) 6 Operations Uvad Thick Mark AsOn Transport — Activity 7 Ready Receiving Allccicd Aitribuks I ruck Driver: Drop, ill 'Noli Ileal Warehouse Infnrnuiion Movx-mem Schedule Activity 7 Operations Find Space Arrange Move men I Aetivit) II Unload I ruck Articled Attribute* Iruck Driver::Rcod) lt> Unload • Transntirt Form Warehouse In Tor million Inventory Information Transport Form Inventory Database -AHivlifI()fwralHw'i Receive hems Determine Place Move Items Mark Transport Request As Perliirmed Update Inventor) Database [ Aciivii) T u l l l Pickup | A lie Clod Attrihirtcs turner:: Arrival turner Information | Invert too Information i * And DoeumcnlutioniiCusioincr :ntor) Database Activity I Arrange Irmnsporl AlTeeied Attribute* ' W are hou*c::Tran spurt Form Inventory Information I ruck Information Transport Schedule:: Warehouse Iranspon orders::Track Driver Activity I Operations Marl Inventory Schedule Trucks •Nulif) Ahoul transport Issue Orders Activit) 1 Record rime A Heeled Atlribtitcs I ruck Driver:Truck Status Iriinspnn In formal ion Activitv 2 Operations Reeord lime Activity I Proceed In pickup AITected Attributes •' Planner:Transport order* Pickup Notification:: Warehouse | Truck S u i us:: Planner Truck Status Ready to kvod::Warehouse Truck S tut u s:: PI an ner Aciivii) I Operations • - Notify annul pickup ' -Report truck status Drive to pickup warehouse • Notify ready lo load • Report truck status Aciivii) 2 Proceed lo dropoff AlTeeied Aiiribuics *W are house ::On Iransnort Dropoff Noti ft cat ion:: Warehouse • ruck Status: :Planncr Truck Status Ready To Unload::Warehouse Truck Status::Planner Aciivii) 2 Operations -Notif) Ahoul Dropoff • Report Truck Stilus Drive To Dropoff -Notify Ready l o Unload • Report Truck Status is and I Xicu mental ki letch liems 4- IK-liscr rum Remove Iter . And DocumenlBlioi s from ihc Sjstcm Figure 3-25 A B P M For The A C M E Warehouse Management Case A full step by step derivation of the diagram can be found in appendix A (including the associated.decomposition for Warehouse), while.the associated agent templates can be found in appendix B. 1 3.8 Summary This chapter introduced the constructs of Ontology-Based Process Modeling. The constructs were then mapped to object-oriented concepts. This led to the introduction of an objected-oriented graphical notation for A B P M diagrams, along.with a procedure for their creation and validation. To illustrate the concepts an example was presented. ( 4. Linking O O E M and A B P M 4.1 Introduction This chapter has the goal of relating Object-Oriented Enterprise Modeling (OOEM) and Activity-Based Process Modeling (OBPM). OOEM is used to describe what things an organization does. ABPM is used to describe how an organization does things. Thus, it is logical that these two methodologies should be related. Also they have similar foundations, which should simplify linking the methodologies. The following diagram illustrates the proposed relationship. Interacting Agents Activity View Full Conversion OOEM View Foundations Partial Conversion OBPM View 7 ~ ~ Foundations Ontological Model Things, properties, etc Figure 4-1. Proposed Relationship And Its Foundations. In particular.we consider .the following as to how we will relate the grammars: • A business process is everything that occurs within the system from the initial request to the system to the final response from the system. • A service is everything that occurs within an object from the initial request to the final response to that request. That is, a process confined to one object (i.e. what is carried out). • An activity is part of a service. An activity is everything within a service confined by (at most) 2 interactions. • An operation is part of an activity. The operations of an activity define how the activity is carried out (and by extension a service is therefore carried out by operations). 65 4.2 Object-Oriented Enterprise Modeling (OOEM) The very first question that comes to mind in systems analysis is why would Enterprise Modeling be needed? The relationship between systems analysis and enterprise modeling can be clarified as: "Systems analysis is the process of understanding the organizational environment and specifying the requirements of it. In order to specify the requirements of a system correctly, a system analyst must first understand the related business areas by developing a model of the enterprise" (Zhao, 1995, citing Gorman 1994). This suggests that the very first activity in systems analysis should'be to develop an enterprise model to describe what an organization does, which will then lead to a process model being developed later to illustrate how an organization does things 4.2.1. O O E M constructs7 This subsection will explain the graphical constructs used in O O E M . To begin there exist two kinds of objects. Those that are external to the system and those that are internal system. Both kinds of objects possess attributes and perform services. External objects make request to objects in the system or have requests made to them from the system. Their services and attributes are not modeled since they are external to the system, and the only thing that matters is the requests they generate or receive. Internal objects receive requests via interface attributes. The request to an object triggers a service in the object. In the course of a service occurring one or more internal attributes may be accessed, as well as one or more requests may be generated. A request may or 7 This discussion of OOEM constructs is based on Wand and Woo 2002 may not have a response generated. And said response may be generated directly or indirectly due to .the service it triggered. . ... . . 4.2.2 Request Propagation External objects affect internal objects by making requests to them. In order to satisfy the request the object may invoke a service. A service is a series of actions that the object performs. During the performance of said actions requests may be generated that affect other objects (or the object itself). The affected objects may then generate requests in fulfilling their responses to the requests made by the first object, and so forth. When all requests to internal objects have been fulfilled and no further requests to internal objects are generated the O O E M is considered complete. Hence an initial request propagates through the system. 4.2.3 How To Generate An O O E M 9 The algorithm for generating an O O E M is as follows: 1. Identify external objects 2. For each external object: 2.1. Identify all requests generated 2.2. For each request, identify: 2.2.1. The object receiving it 2.2.2. The service invoked in the object - • ' .;••;>- "... • -2.2.3. The interface attribute 2.2.4. The response returned by the service (if any) 2.2.5. The internal attributes (if necessary) 2.2.6: Al l requests spawned by the service (if any) and repeat 2.2 for each request 3. If necessary, represent composite/component and/or super/sub-classes using those found in 2.2.1 This discussion of OOEM request propagation is based on Wand and Woo 2002 67 4.2.4 Graphical Representation Of O O E M OOEM has two commonly used representations. The first representation is using an object communication diagram with internal and external object templates. The object communication diagram "...employs simple notation to represent objects (Zhao, 1995)", while the object templates are where the services, attributes, and request information are stored. The second common method is to have all information on the graphic model. For our discussion we will use the second method as illustrated in Figure 4-2 showing an OOEM diagram for the aforementioned ACME Warehouse case (adapted from Wand and Woo, 1999). customer . Pick-up Request Items & Documentation Withdrawal Request Approve/Decline + Reason Office Clerk Withdrawal Requests [Customer Information] Process Withdrawal Request Withdrawal Request Approve/Decline + Reason Transportation Request Warehouse Inventory Information Withdrawal Requests Prepare Loading Requests Prepare Unloading Requests Start Loading Requests Start Unloading Requests Customer Pickup Requests Check Item Availability Process Withdrawal Requests ; Prepare Loading Prepare Unloading Start Loading Start Unloading Process Customer Pickup Requests-' Item Availability llem Existence query Prepare Loading Leaded Stkrt • Prephre Slcrt ; UndmgUnlcadingU'loading . Planner [Transport Planning Requests [Truck Information], Plan Transportation & ^ u c k Assignment Approve/Decline + Reason Transport Orders Transport Status Truck Driver Transportation Requests [Transport Status] Do Transport Orders Legend: /Ex te rna l ObjecK ( Internal Object j Request Response C ) Interface Attribute [Internal Attribute] Service Figure 4-2 O O E M For The A C M E Warehouse Management Case note: Warehouse is a composite object hence the dotted outline ' This discussion of the OOEM algorithm is directly from Wand and Woo 2002 68 4.2.5. A Shortcoming of O O E M In O O E M there is no indication of how the services are performed. A service, is "...a well-defined series of actions taken by the object with the goal of satisfying the request. This series of actions models the state law of the thing" (Zhao, 1995). This sounds good, but it raises the question of where is the 'well-defined series of actions' represented in OOEM? Consider an O O E M service defined simply as sell goods. An executive may be more interested in an expanded view such as stock shelves, price good, display appropriate advertising, and markdown outdated inventory, in order to streamline the selling goods aspect. Also Hui (1997) notes " O O E M describes workflow participants, their responsibilities, and their interactions in a process; it does not capture the execution order of work...." Conveniently, A B P M is concerned with representing series of actions and their order (processes). Hence, if we can find the exact relationship between O O E M and A B P M we can relate what an organization does and how an organization does it in a well-defined manner. . . . . . i , • / • • • • • . 4.3 Basis O f Conversion The consideration to keep in mind during the conversion process is that despite different nomenclatures for the modeling grammars, ontologically a thing is still a thing, a property is still a property, etc. In order to establish a meaningful conversion process between O O E M and A B P M we need to consider how things in each grammar respond to an external event. An external event is a state change of some thing in the system. 69 As illustrated below in figure 4-3 in the enterprise modeling approach an external object generates a request to an object in the system causing it to.gp.from so' to si 1 .The notation s x y denotes the object y in state x. Each time there is a state change it will be noted in the diagram. In servicing the request, Object 1 generates a request to Object 2, and so forth. We can then reinterpret the O O E M approach into a state change view as illustrated in figure 4-4 below. Request s , ' \ s 4 - Request 'External^ ^Object Response s6< Object 1 Response s2< Request Legend S W A Object A is'in state S w after the request is sent S X B Object B is in state S, after the request is received S y B Object B is in state S y after the response is sent S 2 A Object A is in state S 2 after the response is received Request Response s, Figure 4-3 The Enterprise Modelling Approach). v . ; Object 1 Object 2 Object 3 Object 4, External , Object Request Response s 1 Request Request so s 2 S 2 Request Request <5 ' S ' Response - -- - - Figure 4-4 OOEM reinterpreted as a state change view • - - • -• • We can'look at the exact same series of happenings from the process modeling approach. As illustrated below in figure 4-5, an external agent causes a change in Agent 1 causing it 70 to go from so1 to S|'. Each time there is a state change it will be noted in the diagram. In response to the change activity 1 one is triggered. During activity 1 of agent 1 a change is caused in agent 2, and so forth. We can then also reinterpret the A B P M approach into a state change view as illustrated in figure 4-6 below. External Agent Agent 1 Agent 2 Agent 3 Agent 4 1 > Activity 1 •—• Activity 1 — » Activity 1 : v L : s i s s 3 I s 3 Activity 2 ' Activity 1 s , ' s 2 * Activity 3 ""V Legend S , A Agent A is in state S M when activity n begins S r A Agent A is in state S y when activity n ends Agent A Activity n Figure 4-5 The Process Modeling Approach' Object 1 Object 2 Object 3 External Agent Change Change S 1 s 1 S 3 S 1 Change Change Change Change Object 4 s 0 " Figure 4-6 A B P M reinterpreted as a state change view When we take these happenings back to the ontological level basically the same happenings occur in both grammars: As illustrated below in figure 4-7 an external thing modifies some mutual attribute that it shares with some thing internal to the system. 71 Internal thing 1 goes from being in a stable state so1 to an unstable state s/. Internal thing 1 will according to its laws transition from unstable state s/ to stable state S2 1, during which time it will modify some mutual attribute it shares with internal thing 2 causing internal thing 2 to transition from stable state so2 to S | 2 , and so forth. External Thing Mutual Attribute! Modification Internal Event Mutual Attribute Modification Internal Event \ 1 1 1 1 1 1 r~ Internal Thing 1 < s 0 1 , s 1 1 > Internal Thing 1 < s 1 1 , s 2 1 > Internal Thing 2 < s 0 2 , s 1 2 > Internal Thing 2 < s 1 2 , s 2 2 > •table urn table unstable stable stable unstable Mutual Attribute Modification unstable stable Internal Event Mutual Attribute Modification Internal Event . i I i 1 1 I r~ Internal Thing 3 < s 0 3 , s 1 3 > Internal Thing 3 < s 1 3 , s 2 3 > Internal Thing 1 < s 2 1 , s 3 1 > Internal Thing 1 < s 3 1 , s 4 1 > stable unstable unstable stable stable unstable unstable stable Mutual Attribute Modification Internal Event Mutual Attribute Modification Internal Event 1 I - 1 r- 1 Internal Thing 4 Internal Thing 4 < s / , s 2 4 > Internal Thing i Internal Thing 1 < s 5 \ s 6 1 > unstabk, stable stable unstabte unstable stable stable unstable Mutual Attribute Modification External Thing Figure 4-7 The Ontological Meaning Approach Consider the dotted oval in figure 4-5, this could be where there is what we call an O O E M service yet they are distinct A B P M activities. The only way to know is to consider the interactions and what they mean on an ontological level. From the ontological level the main difference between the two grammars is that in A B P M you can expand the internal events and include information that you cannot include in an O O E M . If we map the O O E M and A B P M constructs back to their ontological meanings we get the conversions in table 4-1. OOEM Construct Ontological Model of a process (BWWP) Construct ABPM Construct Object Thing Agent Object Simple Thing Component Agent Composite Object Composite Thing Composite Agent Actor* Agent 72 Non-actor* Resource (An Agent with no services) Actuator* Agent with at least one outgoing interface attribute Propagator* An actuator with at least one outgoing interface attribute affecting an actor Attribute Property Attribute Internal attribute Attribute representing an intrinsic property Internal attribute Interface attribute Attribute representing a mutual property Incoming Interface Attribute + Outgoing Interface Attribute Inherited properties of a subclass Hereditary property- Inherited properties of a subagent Composite object emergent properties Emergent property Composite agent emergent properties Attribute values State Attribute values No services required by an object Stable state No operations required by an Agent Services required by an object Unstable state Operations required by an agent State Change Event State change Execution of a service Internal event Change of an internal attribute, performed by an operation Receiving a request or Receiving a response External event Change of an incoming interface attribute Service Transformation Operation part of a service Activity A sequence of transformations Law Law Law State law State law State Law Transformation law Transformation Law What happens in a system from an initial request to the final response Process* The changes the interacting agents and resources undergo from when one agent becomes unstable to.all agents once again being in stable states Req uest/Response Interaction Change x and y have a shared interface attribute modifiable by x only x acts-on y x has an outgoing interface attribute connected to.an incoming interface attribute of y, only x can modify the outgoing interface attribute and thus by extension the incoming interface attribute of y Definition of a class Functional Schema Definition of an agent Table 4-1 Conversion Table *Denotes a construct from the BWWP ontology not in the BWW Ontology 73 An O O E M transformation deals with only 1 incoming interface attribute and 1 service, whereas the A B P M transformation laws are constraints on two pr more operations hence for this instance there is no mapping of a transformation law in table 4-1. Keep in mind, in an A B P M to O O E M conversion we are moving from a detailed view to a more abstract view. As mentioned above an O O E M service represents a well-defined series of actions, while A B P M delves into what those actions are. Hence we will be losing details in an A B P M to O O E M conversion. Also since an A B P M diagram is a more detailed diagram than an O O E M , it may not be possible to fully construct an A B P M based on an O O E M . Where needed, we will explicitly mention what information needs to be retrieved from domain knowledge that cannot be found in the O O E M diagram. The fact A B P M is more detailed than O O E M also leads to issues of duplication. Duplication is caused in A B P M due to the possibility of the same event being initiated via different activities, which does not occur in O O E M . Consider our solution to the A C M E case from the O O E M perspective, the response from warehouse to office clerk of "Approve/Decline + Reason" is spawned in only 1 service, the service process withdrawal requests. In A B P M the same interaction of notifying the office clerk if the order is approved or declined may occur in the first, third, or fourth activity of the warehouse. 74 4.3 ABPM to OOEM Conversion Due to the fact that an A B P M is more detailed than an O O E M the conversion process may create duplicate requests, responses, or attributes. Duplicates can be left out, since they are not needed in O O E M . We will use the A C M E case as a running example of the conversion process. Table 4-2 illustrates the questions of interest and conversion steps that answer them. Question O f Interest Conversion Step How do we determine the external objects? Step 1. External Object Conversion How do we determine the internal objects? Step 2. Internal Object Conversion How do we determine requests? Step 3. Request/Response Identification How do we determine responses? Step 3. Request/Response Identification How do we determine interface attributes? Step 4. Interface Attribute Conversion How de we determine internal attributes? Step 5. Internal Attribute Conversion How do we'determine services? Step 6. Service Creation Table 4-2 Questions of interest in an ABPM to OOEM Conversion We will now summarize the steps in the A B P M to O O E M conversion process. Following the summarization will be the explanation behind each step and a short example to illustrate each step. 4.4.1 ABPM To OOEM Conversion Steps The Steps for converting from an A B P M to an O O E M are: Step 1. External Object Conversion. Every A B P M external agent becomes an O O E M external object. . , . •. •• Step 2. Internal Object Conversion. Every A B P M anternal agent becomes an O O E M internal object 75 Step 3. Request and Response Identification. Outgoing interface attributes become requests. However if the last outgoing interface attribute is going to the same agent which ^ initiated .the activity in. which, the outgoing interface attribute.is. found the.and the outgoing interface attribute used to initiate the activity'became a request; the outgoing interface attribute becomes a response. This step may encounter duplicate changes, • duplicates can be left out. Step 4. Interface Attribute Conversion. An incoming interface attribute to handle a change from an agent becomes an interface attribute to handle a request. This only applies to those incoming interface attributes of which the associated outgoing interface attribute became a request. Step 5. Internal Attribute Conversion. An A B P M internal attribute becomes an O O E M internal attribute. This step may generate duplicate internal attributes, duplicates can be left out. Step 6. Service Creation. Create a service to process every request. This step requires the modeler to create a service themselves since the service is not present in the process model. 4.4.2 A B P M To O O E M Conversion Step Derivation Step 1. External Object Conversion. Every A B P M external agent becomes an O O E M external object. ' The scope of a system does not change depending on what view of it is used. In our example there is only one external A B P M agent,hence there is only one external O O E M •i object Customer i Office Clerk Warehouse ABPM OOEM Figure 4-8 Demonstrating Step 1 External Object Conversion; ABPM To OOEM Step 2. Internal Object Conversion. Every, A B P M internal agent becomes an O O E M internal object 76 The scope of a system does not change depending on what view of it is used. As well, things of interest are still things of interest regardless of level. In the A C M E case there are four A B P M internal agents hence there will be four O O E M internal objects. Customer : Office Clerk Warehouse i Planner Truck Driver ^) .... .... i: : > . . . . ABPM Office Clerk , Warehouse 1 Planner 1 ruck Driver ' 1 l -1 1 V . O O E M Figure 4-9 Demonstrating Step 2 Internal Object Conversion; ABPM To OOEM Step 3. Request and Response Identification. Outgoing interface attributes become requests. However if the last outgoing interface attribute is going to the same agent which initiated the activity in which the outgoing interface attribute is found the and the outgoing interface attribute used to initiate the activity became a request; the outgoing interface attribute becomes a response10. This step may encounter duplicate changes, duplicates can be left out. According to our mapping in table 4-1 a change of an incoming interface attribute is ontologically equivalent to either a request or a response. An incoming interface attribute is changed via changing the outgoing interface attribute that is associated with it. Duplication may be caused by branching which occurs in process modeling that does not occur in enterprise modeling. This is a complete assumption on our part, made for simplicity. We recognize that an object may receive a response from an object it never made a request to, thus it should be possible to extend this step to be more robust, however we will use this simple form. 77 Customer Office Clerk Warehouse Planner Truck Driver .... C Activity 1 Affected Attributes Custpmer::Withdrawal request Withdrawal Request::Warehouse Order Status: :Customer Activity 1 Operations ABPM ^^Custonier^^ Office Clerk Withdrawal Request 1 Approve/Decline + Reason Warehouse O O E M Figure 4-10 Demonstrating Step 3 Request and Response Identification; ABPM To OOEM *Recall a request appears at the tail of an arrow, a response at the head of an arrow Step 4. Interface Attribute Conversion. An incoming interface attribute to handle a change from an agent becomes an interface attribute to handle a request. This only applies to those incoming interface attributes of which the associated outgoing interface attribute became a request. An incoming interface attribute is how an agent handles (i.e. receives) external events. An external'event is handled (received) by an object through an interface attribute. For those incoming interface attributes which the associated outgoing interface attributes became aresponse an interface attribute is not needed in the object since responses do not have interface attributes.associated with them.11., . , :. " Responses can actually have interface attributes to handle them, however due to convention they typically do not. 78 Customer Office Clerk Warehouse Planner Truck Driver Activity 1 Affected Attributes Withdrawal Request::Warehouse Activity 1 Operations ^ Activity 1 Affected Attributes Office Clerk::Withdrawal Request - • Activity I Operations ABPM Office Clerk Withdrawal Request I 1 | Withdrawal Request k-^ 1 OOEM Figure 4-11 Demonstrating Step 4 Interface Attribute Conversion; ABPM To OOEM Step 5. Internal Attribute Conversion. A n A B P M internal attribute becomes an O O E M internal attribute. This step may generate duplicate internal attributes, duplicates can be left out. Accord ing to our mapping in table 4 - 1 : an internal attribute is the same in both grammars. The duplication is caused by branching that is present in process models but not in enterprise m o d e l i n g : " " -Customer i Office Clerk" Warehouse: Planner Truck Driver .... 1: y Activity 1 Affected Attributes -Authorization Status, . • j . . i : Activity 1 Operations \ Office Clerk [Authorization Status] ABPM OOEM Figure 4-12 Demonstrating Step 5 Internal Attribute Conversion; ABPM To OOEM 79 Step 6. Service Creation. Create a service to process every request. This step requires the modeler to create a service themselves since the service is not present in the process model. The important thing to keep in mind in that in O O E M every request requires a service. What we know is that every request started as an outgoing interface attribute. If there are duplicate outgoing interface attributes we know the activities in which the outgoing interface attribute is found are part o f the same service, since a request is spawned by only one service. Thus a potential candidate for a service is the operation that changes the outgoing interface attribute, and when there are duplicate outgoing interface attributes there are multiple candidate operations for a service. It is stil l a modeler's decision to either use one o f the candidate operations as a service or to create a service from scratch. Customer X withdrawal Request Customer\ withdrawal Request Office Clerk Withdrawal Requests [Authorization Status] O O E M Before Step 6 Office Clerk Withdrawal Requests [Authorization Status] Process Withdrawal Request Warehouse ^ v J Warehouse • O O E M After Step 6 Figure 4-13 Demonstrating Step 6 Service Creation; A B P M T<> O O E M A t this point we are finished the conversion. If we compare our solution after the conversion to that in section 4.2.4 above we see the difference is the post conversion solution has more internal attributes. Since internal attributes are optional in O O E M the solution is the same. 80 4.5 OOEM to ABPM Conversion The following questions are of interest in an A B P M to O O E M conversion: Question Of Interest Conversion Step How do we determine the external agents? Step 1. External Agent Conversion How do we determine the internal agents Step 2. Internal Agent Conversion How do we determine activities? Step 5. Activity Creation How do we determine outgoing interface attributes? Step 3. Outgoing Interface Attribute Identification How do we determine incoming interface attributes? Step 4. Incoming Interface Attribute Identification How de we determine internal attributes? Step 7. Internal Attribute Identification How do we determine operations? Step 8. Operation Identification. How do we determine state laws? Step 5. Activity Creation How do we determine transformation laws? Step 5. Activity Creation Step 8. Operation Identification How do we determine resources? Step 9. Resource Identification Table 4-3 Questions of interest in an OOEM to ABPM Conversion We will now summarize the steps in the O O E M to A B P M conversion process. Following the summarization will be the explanation behind each step and a short example to illustrate each step. 4.5.1 OOEM To ABPM Conversion Steps The Steps for converting from an O O E M to an A B P M are: Step 1. External Agent Conversion. Every O O E M external object becomes an external A B P M agent. •• . • . . : Step 2. Every . O O E M internal object becomes an A B P M internal agent Step 3. Outgoing Interface Attribute Identification. A l l requests and responses become outgoing.interface attributes. • • Step 4. Incoming Interface Attribute Identification. A l l outgoing interface attributes identified in step 3 require an incoming interface attribute. As well when we create the incoming interface attribute we need to show the agents share a mutual attribute. Step 5. Activity Creation. At this point we can create activities. Each time there is an incoming interface attribute without a transformation law requiring other incoming 81 interface attributes to change to start the operations of the activity, it is the beginning of a new activity. We can create transformation and state laws for the incoming interface attributes using domain knowledge. Step 6. Outgoing Interface Attribute Assignment. We can now assign the outgoing interface attributes to the appropriate activity using domain knowledge. Duplicate assignments are allowed. Step 7. Internal Attribute Identification. Every OOEM internal attribute becomes an ABPM internal attribute. We can the assign them to the appropriate activity using domain knowledge. Additional internal attributes may need to be created using domain knowledge. Duplication may occur Step 8. Operation Identification. Create the operations that will be invoked to change the outgoing interface and internal attributes using domain knowledge. We also have to create transformation laws using domain knowledge that show if there are any restrictions on the operations of the activity. Step 9. Resource Identification. In the event there are agents only have incoming interface attributes in all of their activities they are resources and need to be converted to the appropriate resource notation. 4.5.2 O O E M To A B P M Conversion Step Derivations Step 1. External Agent Conversion. Every OOEM external object becomes an external ABPM agent. The scope of a system does not change depending on what view of it'is used. In our example there is only one external OOEM object, hence there is only one external ABPM agent. '.v.. ... ... Customer OOEM ABPM Figure 4-14 Demonstrating Step 1 External Agent Conversion; OOEM To ABPM 82 Step 2. Internal Agent Conversion. Every O O E M internal object becomes an A B P M internal agent The scope of a system does not change depending on what view of it is used. As well, things of interest are still things of interest regardless of level. In the A C M E case there are four O O E M internal objects hence there will be four A B P M internal agents. Warehouse ^ Office Clerk 1 \ . Planner Truck Driver O O E M Customer Office Clerk Warehouse Planner Truck Driver A B P M Figure 4-15 Demonstrating Step 2 Internal Agent Conversion; OOEM To ABPM Step 3. Outgoing Interface Attribute Identification. All requests and responses become outgoing interface attributes. Since requests and responses model interaction between two things in O O E M , they need to model interaction between two things in A B P M . A change models interaction in A B P M . A change is the initiation of an event external to ;an agent.-External events change incoming interface attributes: Since incoming interface attributes are only changed when the associated outgoing interface attributes are changed by operations in other agents, requests and responses must become outgoing interface attributes. 83 1 Office Clerk Withdrawal Request I 1 Approve/Decline ' + Reason \ V Warehouse • 1 v . O O E M Customer Office Clerk Warehouse : Planner Truck Driver Order Status::Customer Withdrawal Request::Warehouse i ABPM Figure 4-16 Demonstrating Step 3 Outgoing Interface Attribute Identification; OOEM To ABPM Step 4. Incoming Interface Attribute Identification. All outgoing interface attributes identified in step 3 require an incoming interface attribute. As well when we create the incoming interface attribute we need to show the agents share a mutual attribute. / — By definition all outgoing interface attributes need an associated incoming interface attribute. We do not need additional information from the O O E M for this step. Customer j Office Clerk Warehouse Planner Truck Driver : Order Status::Customer . . : Withdrawal Rcqucst::Warehouse A B P M , before step 4 Customer Office Clerk Warehouse Planner Truck Driver Office Clerk::Ordcr Status *— -Order Status::Customer : Withdrawal Request::Warehouse-i—"Office Clerk::Withdrawal Request A B P M , after step 4 Figure 4-17 Demonstrating Step 4 Incoming Interface Attribute Identification; OOEM To ABPM Step 5. Activity Creation. At this point we can create activities. Each time there is an incoming interface attribute without a transformation law requiring other incoming 84 interface attributes to change to start the operations of the activity, it is the beginning of a new activity. We can create transformation and state laws for the incoming interface attributes using domain knowledge. As noted in our mapping in table 4-1 the transformation law information is not present in O O E M . Thus we will have to use our domain knowledge. At this point (in the A C M E case) there do not appear to be any transformation laws. That means every incoming interface attribute will be the beginning of a new activity. We need to create the state laws using domain knowledge since they are not represented directly in O O E M Customer Office Clerk Warehouse Planner Truck Driver Customer::Withdrawal Request Withdrawal Rcquest.:Warchouse Order Status: :Customer ABPM, before ste p5 • Customer Office Clerk Warehouse; Planner Truck Driver Withdrawal Request::Warehouse Order Status-Customer : Activity I Withdrawal Request : : I | Affected Attributes j j j | Customer-Withdrawal Request j j j ' '." . •" ' j lAclivitv I Operations I j I i ABPM, after step 5 Figure 4-18 Demonstrating Step 5 Activity Creation; O O E M To A B P M *note the attributes above the activity are outgoing interface attribute that have not yet been - 1 assigned to an activity ! : . • Step 6. Outgoing Interface Attribute Assignment. We can,now assign the outgoing interface attributes to the appropriate activity using domain knowledge. Duplicate assignments are allowed. 85 Duplicate assignments can occur since the same transformation may be triggered in different activities. We must use the domain knowledge to assign the outgoing interface attributes to the appropriate activity since the expanded activity view is not present in O O E M . Customer Office Clerk Withdrawal Request::Warehouse Order Status:;Customcr , Activity 1 Withdrawal Request jAffcctcd Attributes |Customer::Withdrawal Request Activity 1 Operations Activity 2 Notify about order status Affected Attributes Warehouse: :Order Status Activity 2 Operations Warehouse : Planner i Truck Driver ABPM, before step 6 Customer Office Clerk Activity I Withdrawal Request Affected Attributes |Customcr::Withdrawal Request Withdrawal Request Warehouse lOrdcr Status::Customer Activity I Operations Activity 2 Notify about order status Affected Attributes Warehousc::Ordcr Status lOrdcr Status::Customcr Activity 2 Operations Warehouse : Planner : Truck Driver ABPM, after step 6 Figure 4-19 Demonstrating Step 6 Outgoing Interface Attribute Assignment; OOEM To ABPM Step 7. Internal Attribute Identification. Every O O E M internal attribute becomes an A B P M internal attribute. We can the assign them to the appropriate activity using domain knowledge. Additional internal attributes may need to be created using domain knowledge. Duplication may occur According to our mapping an internal attribute is the same in both grammars. However in O O E M internal attributes are optional, thus not all the A B P M internal attributes may be identified. A modeler will have to use their domain knowledge (in our case that's the A C M E Case) to identify any other internal attributes. 86 Planner [Truck Information] I - J O O E M Customer j Office Clerk Warehouse Planner Truck Driver Truck Information Inventory Information* Transport Information* ABPM Figure 4-20 Demonstrating Step 7 Internal Attribute Identification; O O E M To A B P M * denotes an attribute created using domain knowledge Step 8. Operation Identification. Create the operations that w i l l be invoked to change the outgoing interface and internal attributes using domain knowledge. We also have to create transformation laws using domain knowledge that show i f there are any restrictions on the operations o f the activity. We cannot use O O E M services to create A B P M operations since as mentioned above a service may encompass many operations. The only way outgoing interface and internal attributes are changed is via operations. A s noted in our mapping in table 4-1 the transformation law information is not present in O O E M . Thus we w i l l have to use our domain knowledge. , . . 87 Customer Office Clerk i Warehouse : Planner Activity I Withdrawal Request Affected Attributes Customer::Withdrawal request Authorization Status Withdrawal RequestrWarehouse Order Status::Customer Activity 1 Operations Truck Driver A B P M , before step 8 Customer : Office Clerk Activity I Withdrawal Request Affected Attributes Customer::Withdrawal request Authorization Status Withdrawal Rcquest:Warehouse Order Status: :Customer Activity I Operations Contact Warehouse and Notify About Status are mutually exclusive Authority Check Contact Warehouse Notify about status Warehouse : Planner : Truck Driver A B P M , after step 8 Figure 4-21 Demonstrating Step 8 Operation Identification; OOEM To ABPM Step 9. Resource Identification. In the event there are agents only have incoming interface attributes in all of their activities they are resources and need to be converted to the appropriate resource notation. Resources are the only things that can receive change that are not agents. Note: for our example there are no outgoing interface attributes that meet the criteria of step 9, however an example would be something like figure 4-22 below. Customer Office Clerk : Warehouse Usc:;Stapler Staple Stuff" Planner Truck Driver Stapler Activity I staple Affected Attributes Office Clerk::Use ABPM, before step 9 Customer Office Clerk ; Warehouse Use:: Stapler Staple Stuff" Planner : Truck Driver Stapler Office Clerk::Use A B P M , after step 9 Figure 4-22Demonstrating Step 9 Resource Identification; OOEM To ABPM 88 4.6 Converting Decompositions and Compositions It should be noted that in our running conversion examples above we neglected to mention and demonstrate the warehouse decomposition. The reason behind that is there is no need to show it. The Steps apply the same way in the decomposed view or the composite view since in the decomposed view the component things are still things that interact and will still interact in either O O E M or A B P M , the difference is the information included. 4.7 Converting Subagents/Superagents And Subclasses/Superclasses It should also be noted that in our running conversion examples above we neglected to mention and demonstrate the generalization/specialization conversions between agents and classes and vice versa. They are still things so thus the steps apply the same way to the things in these relationships. The modeler needs to ensure that the correct generalization and specialization notation for the grammar is used post conversion. 4.8 Summary This chapter explained O O E M . It delved into the relationship between O O E M and A B P M and introduced means to convert form one view to the other. This chapter also used the examples from the A C M E case to demonstrate the conversion process. 89 5. Design Principles of an ABPM CASE Tool 5.1 Introduction One of the goals of this thesis is to take OBPM from an algorithm to a graphical modeling grammar. To make it more useful as a modeling grammar a tool that supports the A B P M modeling process can be developed. This chapter presents a set of design principles for taking A B P M a step further and implementing it as a C A S E tool. A B P M has very specific semantics that proper design of a C A S E tool can take advantage of. Our only assumption is that there already exists some sort of C A S E tool upon which O O E M has been implemented. 5.2 System Goals The main goal of the C A S E tool development is to implement A B P M in a visual manner. This manner should be consistent with both the semantics and the activity-based graphical representations presented back in chapter 3. However as a secondary goal, it should also support the A B P M to O O E M conversion and O O E M to A B P M conversion from chapter 4. 5.3 System Requirements Based on Zhang's requirements for a visual O O E M C A S E tool (Zhang, 1998), we can say the C A S E tool will be required to support the entire A B P M model creation process. In the initial stage the tool should allow the process modeler to gather and organize information into an understandable model. The tool should then be capable of applying the A B P M rules to a developed model to evaluate its semantic correctness. Once 90 evaluated it should be possible to change the model as needed until the rules are met. Finally the tool should allow for the modeler to suspend activity at any time and resume it later. No information should be lost during these actions (Zhang, 1998). Also, the case tool should be able to support the analysis stage. That is, the user should be able to change elements as desired so that they can see what happens when changes are made. (Zhang, 1998) This directly supports process redesign efforts. The C A S E tool should also support the model conversion process from A B P M to O O E M and from O O E M to A B P M without losing any information during the conversion process. In order to meet these requirements we need to consider two main areas: what constructs to represent in the C A S E tool and what interactions a user will have with the C A S E tool. 5 . 4 Constructs to Represent If we want to build upon the existing O O E M C A S E tool then when it comes to constructs to represent we can first need to consider what is already present in an O O E M C A S E tool. We can then compare that with the constructs in A B P M . After the comparision we should then know what needs to be added to the C A S E tool. The basis of comparison will be the respective metamodels of both grammars. The O O E M metamodel, as adapted from Tan (Tan, 1997) and his work on an O O E M C A S E 91 tool to show attributes are either internal or interface and responses can be spawned by services, is presented below in figure 5-1. Spawn Figure 5-1 The O O E M Metamodel (adapted from Tan 1997) We developed the A B P M metamodel back in chapter 3. The question now is how do we relate the metamodels to create a C A S E tool metamodel? The question is answered the same way as how we related the two modeling grammars, use their ontological foundations to establish what is there and what is not. Consider figure 5-2 below, as an O O E M Metamodel with the ontological meanings of the constructs added. 92 0+ Receive 11 Object Thing Response Interaction 1 + 0+ 0+ Generate — Sent " 1+ 1 + Generalization / Specialization of Objects External External Thing Attribute Attribute I 1 + Request Interaction I 1 1 Interface Internal Internal Thing Generalization / Specialization Internal Internal Attribute 1 + Invoke Interface Mutual Attribute Provide I— A c c e s s o+ May-Own o+ May-Provide Spawn 1+ o+ Service 0-H Transformation Spawn Figure 5-2 O O E M Metamodel with ontological meaning included When we look at the ABPM model with the ontological meanings of the constructs added we get the diagram in figure 5 - 3 below. Since an activity has no direct ontological mapping we did not include its ontological meaning. The boxes that are double-lined denote the constructs that are directly equivalent,, and are thus already represented. The dotted box denotes that although not directly equivalent on their own, combined an outgoing interface attribute and incoming interface attribute have an ontological meaning of interaction. 93 1 A g e n t Thing C h a n g e o n l y h a v e (jeneraluation / Specialization E x t e r n a l A g e n t External Thing Internal A g e n t Internal Thing Operations 1+ Transformation 11 Restrict ! | n c d m j r r g : i h ( e j f a c e : ; : • Y acted, upon by X. O u t g o i n g ! ; •(nlerfatee:-X acts, upon Y S t a t e L a w State Law T r a n s f o r m a t i o n L a w Transformation Lawl Figure 5-3 ABPM metamodel with ontological meaning included From the two metamodels we can thus see, an A B P M includes all the constructs presented in an O O E M . However the A B P M has additional constructs not present. Thus the additional constructs to be represented are: Resource, Activity, Laws, State Laws, and Transformation Laws, While the constructs that are already represented (ontologically) but need to be configured to accept input as A B P M constructs are: Agent, External Agent, Internal Agent, Attribute, Internal Attribute, Operations, Incoming Interface Attributes, and Outgoing Interface Attributes.. 94 5.5 User Interactions The user interactions for the A B P M C A S E tool are those related to drawing A B P M diagrams, converting A B P M and O O E M diagrams, display option, and semantic checker. 5.5.1 Drawing ABPM Diagram Interactions For all elements a user should be able to add, delete, and modify as desired. For the elements where it is not obvious how to add, delete, or modify it will be discussed below. 5.5.1.1 Internal and External Agents We recommend that when the agents and resources are implemented, that each construct will have the associated agent template implemented with it as well. This will allow the user to work from the view that is more convenient for them. When the user wishes to make a modification they should merely have to specify which view they wish to work in. If chosen, changes made in the agent template view should be reflected in the diagram when the template is closed. 5.5.1.2 Agents Sharing A Mutual Attribute When an incoming interface attribute has an outgoing interface attribute associated with it, an arrow from the operation that changes the outgoing interface attribute to the incoming interface attribute should drawn. In the event the same outgoing interface attribute occurs in more than one activity, then when they are associated with the incoming interface attribute a function will have to run that automatically generates a double-lined arrow. When compressed without operations the arrows can automatically 95 be redrawn to connect directly from the outgoing interface attributes to the incoming interface attributes. Mores is said about compression in section 5.5.2 below. 5.5.1.3 Superagents And Subagents If it is decided to turn an agent into a superagent, nothing special happens. When it is decided to create a superagent from existing (sub)agents. A function to automatically extract the agent activity template of the superagent should be called. The function would pull out the attributes and operations of each subagent that they have in common. A notification (reminder) that a subagent needs to have additional attributes not present in the superagent may need to occur 1 3 . A n y changes that are now routed to the superagent should automatically be rerouted from the subagents in the diagram. The third possibil ity is that existing agents could be grouped such that one o f them is set to be a superagent and the rest are subagents. In this event the superagent would need to have its agent activity template updated, much like in the aforementioned second possibil ity, from the common attributes and operations of the subagents. The original properties o f the superagent and those added from the subagent(s) w i l l need to be organized into activities by the user. A prompt should be issued to remind the user to reorganize the properties o f the super agent. A l l attributes and operations of the superagent that exist in the subagents should be removed. The attributes and operations o f the subagent then need to be reorganized into activities. A prompt ( i f necessary) should be issued to remind a user that 1 3 This reminder may also help identify typos made in the subagents preventing a superagent from having a non-empty agent template. 96 a superagent cannot have an empty agent activity template and that each subagent must have its own attributes and operations that are not present in the superagent. 5.5.1.4 C o m p o s i t e a n d C o m p o n e n t A g e n t s A user should be able to switch between the composite and components view of an agent at any point they desire. There are three possibilities with composite and component agents. The first possibility is there is a composite agent that is decomposed down into its components. In this case a prompt is issued to the user for how many components there are and what there names will be. Since we already know what the incoming and outgoing changes to the composite agent are we can issue a prompt for both which incoming changes (actually the interface attribute associated with the incoming change) go to which agent, then the entire activity is assigned to the component. The last prompt necessary is for what changes (if any) do the components use to interact with each other. The user then needs a reminder the component interactions all require activities. The second case is that several components are combined to create a composite object. In this case the agent activity template of the composite is populated from the agent templates of the components. Any interactions between the components are left out. When the components have the same activity it is only entered in the composite once. When the components have an activity that is the same except for the incoming interface attributes then only one activity will be entered in the composite, but both incoming interface attributes will be placed in the activity and the user will be prompted for the 97 transformation law that exists. The user should then be prompted for the emergent attributes and operations of the composite. Lastly, the user.should be asked if they wish to view the composite in the composite view or the decomposed view in the diagram. The last case is that from a collection of agents one is decided to be a composite agent and the rest component agents. In that event the agent activity template of the composite agent needs to be altered to: show its attributes and operations are emergent and to include the attributes, and operations of the component agents (as aforementioned in creating a composite) and to denote which component agent contributed which attribute and operation. If a component and the composite agent have the same attribute or operation it is denoted as being contributed from the component not the composite. Lastly, the user should be asked if they wish to view the composite in the composite view or the decomposed view in the diagram. 5.5.2 Display Option A display option should be included to allow the A B P M to be shown as is, or in any compressed view the user desires. Since the information is stored in the agent templates viewing the compressed agents will not lose any information. Specifically, the user should be able to choose the display method for any agent in the model. This can be done for various reasons such some large agents may need to be compressed in order to capture the entire system in a confined area (e.g an 8.5"xl 1" page), a user may only be interested in viewing the interactions, etc. 98 5.5.3 O O E M And ABPM Conversions As established in chapter 4 it is possible to convert an O O E M to an A B P M and vice versa. We assume there already exists some sort of C A S E tool which has implemented O O E M . The conversion process should be automated as much as possible to prevent human error. However at steps where there is the possibility of ambiguity or lack of information a user could be prompted to for the appropriate information. Each model should have its own window. As well, the source file used to create the second diagram should not be changed by the conversion process. 5.5.3.1 ABPM to O O E M Converter For the sake of convenience we will assume the user has developed an A B P M that has passed a semantic integrity check. The starting point is the user has selected the option to convert the A B P M to an O O E M . The C A S E tool should initialize a new O O E M window that is linked to the current A B P M . Using the A B P M to O O E M conversion process from chapter 4, we will discuss the conversion steps and highlight those that require user intervention. Step 1. External Object Conversion. Every A B P M external agent becomes an O O E M external object. This is a step that can be automated. The system can automatically create a new empty external object template for each external object. As well it can create the associated graphical construct for an external object and place it in the diagram. Step 2. Internal Object Conversion. Every A B P M internal agent becomes an O O E M internal object 99 This is a step that can be automated. The system can automatically create a new empty internal object template for each external object. As well it can create the associated graphical construct for an internal object and place it in the diagram. Step 3. Request and Response Identification. Outgoing interface attributes become requests. However if the last outgoing interface attribute is going to the same agent which initiated the activity in which the outgoing interface attribute is found the and the outgoing interface attribute used to initiate the activity became a request; the outgoing interface attribute becomes a response. This step may encounter duplicate changes, duplicates can be left out. This step can be automated. The system can first eliminate duplicate changes. Then it can convert then to requests or responses as dictated. The requests and responses can then be automatically placed in the diagram. Step 4. Interface Attribute Conversion. An incoming interface attribute to handle a change from an agent becomes an interface attribute to handlea request. This only applies to those incoming interface attributes of which the associated outgoing interface attribute became a request. This step can be automated. By keeping track of the changes that became requests in the previous step'the system can know which incoming interface attributes to convert to interface attributes.The interface attributes can the be automatically placed in the' diagram. Step 5. Internal Attribute Conversion. An A B P M internal attribute becomes an O O E M internal attribute. This step may generate duplicate internal attributes, duplicates can be left out. • 1 This step can be automated. The system can first eliminate duplicate attributes. Then it can put the internal attributes into the object template. Then the internal attributes can automatically be placed in the diagram. 100 Step 6. Service Creation. Create a service to process every request. This step requires the modeler to create a service themselves since the service is not present in the process model. The system can automatically place one of the operations that are candidates for a service in the diagram. The user can reminded that the services may not be correct, and that they may either need to use a different one of the candidates for a service or create a new service altogether for each request. 5.5.3.2 O O E M to A B P M Converter For the sake of convenience we will assume the user has developed an O O E M that has passed a semantic integrity check. The starting point is the user has selected the option to convert the O O E M to an A B P M . The C A S E tool should initialize a new A B P M window that is linked to the current O O E M . Using the O O E M to A B P M conversion process from chapter 4, we will discuss the conversion steps and highlight those that require user intervention. Step 1. External Agent Conversion. Every O O E M external object becomes an external A B P M agent. This is a step that can be automated. The system can automatically create a new empty external agent template for each external agent. As well it can create the associated graphical construct for an external agent and place it in the diagram. 101 Step 2. Internal Agent Conversion. Every O O E M internal object becomes an A B P M internal agent This is a step that can be automated. The system can automatically create a new empty internal agent template for each internal agent. As well it can create the associated graphical construct for an internal agent and place it in the diagram. Step 3. Outgoing Interface Attribute Identification. All requests and responses become outgoing interface attributes. This is a step that can be automated. The system can do the conversion and then put them into the first activity of the agent template to temporarily store them. Step 4. Incoming Interface Attribute Identification. Al l outgoing interface attributes identified in step 3 require an incoming interface attribute. As well when we create the incoming interface attribute we need to show the agents share a mutual attribute.' This step can be automated. The outgoing interface attributes can be dissected for the 'attribute changed' part of the incoming interface attribute, while the agent possessing the outgoing interface attribute is the 'agent doing the change' part. The incoming interface attribute can then be stored in the agent template of the agent that possesses it. Step 5. Activity Creation. At this point we can create activities. Each time there is an incoming interface attribute without a transformation law requiring other incoming interface attributes to change to start the operations of the activity, it is the beginning of a new activity. We can create transformation and state laws for the incoming interface attributes using domain knowledge. The system can automatically put each incoming interface attribute into a new activity. The user can then be prompted for any transformation laws that exist and what incoming interface attributes belong'to the same activity. 102 Step 6. Outgoing Interface Attribute Assignment. We can now assign the outgoing interface attributes to the appropriate activity using domain knowledge. Duplicate assignments are allowed. This is a manual process (unless there is only one activity in which case the assignment can be automatically done done), in which the user can be prompted for which activity the which outgoing interface attributes is altered in. Step 7. Internal Attribute Identification. Every O O E M internal attribute becomes an A B P M internal attribute. We can the assign them to the appropriate activity using domain knowledge. Additional internal attributes may need to be created using domain knowledge. Duplication may occur This step can be partially automated. The system can do the conversion for the existing O O E M internal attributes, however the user will have to be prompted for any extra attributes. The user also needs to be prompted for which activities the attributes are used in. ' ! <••••••" " • ' - ' ' •' ••• ' ' • Step 8. Operation Identification. Create the operations that'will be invoked to change the outgoing interface and internal attributes using domain knowledge. We also have to create transformation laws using domain knowledge that show if there are any restrictions on the operations of the activity. It would drive a user insane if they were prompted for an operation for every outgoing interface and internal attribute. We suggest one reminder about the fact the user needs to create operations for every outgoing interface and internal attribute. At this point the diagram would be in a compressed view that has no operations. The user can then rely on the semantic checker (discussed in the next section) to ensure they have an operation for every outgoing and interface attribute. 103 Step 9. Resource Identification. In the event there are agents only have incoming interface attributes in all of their activities they are resources and need to be converted to the appropriate resource notation. This is a step that can be automated. The system can check if there are any agents with that only have incoming interface attributes in its activities. If there are it can automatically convert the agent to a resource. 5.5.4 Semantic Checker The requirement of a semantic checker is rather intuitive. Modern tools (for example Microsoft Word) provide facilities that perform error checking for the user. This makes the tool more useful to the user. As well a tool that can check whether an A B P M diagram is correct will help to reduce errors and lead to better models being created. How are errors introduced to an ABPM? Errors are introduced byviblating the semantic integrity of the language, that is, by violating the rules. Hence a semantic checker will be a useful part of the tool to ensure the semantic integrity rules are followed. When a rule is violated the error can be highlighted for the user. The rules (from chapter 3 ) and how they should be implemented in the system are below: 1. Every change in the system should be able to be traced back to an initial change from an external agent. The incoming interface attribute for each activity can be back tracked to the outgoing interface attribute. The system can then backtrack the incoming interface attribute of the activity that the outgoinginterface is a part of until an external agent is reached. If an external agent is not reached then there is an error. 2. Every agent must have at least one activity. The system can check if the activity count for each agent is greater than or equal to one. If it is not, there is an error. 3. Every Activity must have at least one operation The system can check if the operation count for each activity is greater than or equal to one. If it is not, there is an error. 4. Every resource must only have incoming interface attributes. The system can check each entry in the resource template. If they are not incoming interface attributes (notationally in the form: agent doing the change:attribute changed), there is an error. 5. Every attribute is changed by one operation. In the agent template every outgoing interface attribute and interface attribute should have one and only one operation. The system can check if this true, if not there is an error. ' 6. Every activity can only have one incoming interface attribute unless governed by a transformation law. In the agent template if the transformation law associated with an incoming interface attribute is blank there should only be one incoming interface attribute for each activity. The system can check if this is true, if not there is an error. 7. Every outgoing interface attribute must have a corresponding interface attribute and vice versa. i 105 Two checks occur here. For each outgoing interface attribute, there has to be a check in the activities of the agent it lists the on the 'agent being changed' part of the outgoing interface attribute that there is in fact an incoming interface attribute for it. If not there is an error. This check should keep track of the incoming interface attributes it found, if there are other incoming interface attributes not found in during the check then there is an error since those incoming interface attributes do not have an outgoing interface attribute associated with them. 8. Every composite agent must possess emergent attributes and operations not present in the component agents The system can check if there additional attributes and operations not present in the components. IfThere are not any then there is an error 9. Every subagent must possess attributes and operations that are unique to the subagent and are not inherited from the superagent. The system can check if there additional attributes and operations not present in the superagents. If there are not any then there is an error 5.6 Design Limitations The first implementation limitation is that the proposed design guidelines use the notation developed in chapter 3. A user may wish to develop an entirely different notation for the constructs. Even if that is the case, the requirements developed previously will still hold. They merely need to be implemented using the new notation. The second limitation is that A B P M itself has not yet been rigorously tested, thus errors in A B P M will have propagated through the design. The third limitation'is that the A B P M C A S E tool design 106 was based partially on the design of an existing OOEM CASE tool. This may have introduced flaws of its own since an OOEM lacks information relative to ABPM. 5.7 Summary This chapter proposed design principles for ABPM as a CASE tool. It defined the constructs that needed to be added to an OOEM CASE tool. It also discussed user interactions with such a CASE tool. The last thing this chapter considered was possible design limitations. 107 6 Conclusions And Future Research 6.1 Thesis Summary This thesis presented a new method for modeling organizational processes; Object-Oriented Activity-Based Process Modeling (ABPM). We first looked at business process modeling in general and why improvements were needed. We then proceeded to further develop a modeling algorithm proposed by Wang (Wang 2002) into a graphical modeling grammar. We created the grammar by combining the constructs of the Ontology-Based Process Modeling (OBPM) algorithm with object oriented constructs. We then tied the new A B P M constructs to specific graphical constructs. The essential constructs in A B P M are agent, attribute, and operation. We then developed a modeling process for using the graphical A B P M constructs. The modeling process was based partially on the work by Wang (Wang 2002) and partially on work by Wand and Woo (Wand and Woo, 2002). The modeling process is based upon rules for: model scope, agent identification, operation inclusion, attribute inclusion, attribute ownership, composite agents, sub-classification, and new activities. We also include a set of semantic integrity rules that can be used to check if a created A B P M is semantically correct. Based upon their similar theoretical foundations and purposes we then delved into the relationship between Object-Oriented Enterprise Modeling (OOEM) and A B P M . We developed a means to convert from one model to another, noting along the way that A B P M contains more information than O O E M which should be the case since A B P M is a more detailed view of organizational activity. At this point we created design principles for the implementation of A B P M as a C A S E tool. We discussed the functional and non-functional requirements of such a C A S E tool. Next we proposed a possible development platform for A B P M . We also considered the possible limitations of the proposed design architecture. 6.2 Contributions This thesis developed the OBPM algorithm into an objected graphical modeling language and process. The A B P M constructs have specific and well-defined semantics for real world business process representation. A noteworthy contribution is the refinement of the change propagation algorithm which is based upon a set of ontologically derived rules to create a systematic process for modeling a business process. The strength of the algorithm is from its ontological real world foundations rather than programming or data design rules of thumb. This thesis also makes a contribution by exploring the relationship of A B P M and O O E M . Both languages are designed to model a specific view of organizational activity, irrespective of how a later information system artifact will be built. By relating the two grammars using ontological foundations we can acquire greater understanding of an organization without losing information. 109 Our last contribution the development of a design principles for an A B P M C A S E tool that is implementation independent means that no matter how one decides to implement A B P M if they follow our requirements they will be able to create a tool to fully support the business process model generation process. 6.3 Limitations And Future Research The foundations of A B P M were established by previous research. However A B P M itself is a new grammar and methodology. It has been applied to very few cases and testing so far has been limited. It also needs to be tested beyond an academic setting and in the real world. This will allow for validation of the grammar and method. We also make a simplifying assumption for our A B P M to O O E M conversion process. During step 3 (request and response identification), regarding how to identify responses, we assume that an object can never receive a response from an object that has never made a request to it. Future research can be done on the ontological nature of both requests and responses, so that a better form of the request and response identification step during the A B P M to O O E M conversion process can be developed. Another possible limitation is that we talk about both analysis principles (e.g. operation, activity) and design principles (e.g. attributes) in the same grammar. This is due to the fact that we use ontological principles that have no business meaning, with a goal of representing processes. Therefore when using the grammar for analysis and/or design we 110 may have to convert ontological concepts to business concepts. For example attributes may need to be converted to documents. Further research could be done to determine the extent of the conversions necessary, if any. Other areas of future research could focus on: • A B P M does not allow for new state variables. The model A B P M is based upon does not allow for creating and eliminating things. Al l changes are modeled as changes of attribute values. Hence A B P M does not allow for creating and eliminating resources. There are two possible responses to this observation. First we can analyze everything in the domain, if all resources have been identified beforehand, they cannot be changed if they do not exist. Second, ontologically nothing appears or disappears, we just change its name. Research could be undertaken to determine which would be the more appropriate manner to deal with new or disappearing state variables. • Data is disregarded is O B P M and thus by extension in A B P M . However computerized information systems tend to primarily pass data. An ontologically based data modeling grammar and method that is related back to A B P M would fill this gap. • Wand, Woo, and Jung (Wand, Woo, and Jung, 2000) proposed a means to convert from an O O E M to a logical design of an information system. It should be possible to convert from an A B P M to a logical design of an information system since we can convert an A B P M to an O O E M and then a logical design of an information system. The question arises is it possible for a direct conversion from A B P M to a logical design? Another question also arises concerning if it is possible to create a logical directly design from an ABPM will it be the same as the logical design from an OOEM, and if they are different which one is better? • Although we have proposed a set of design principles for a CASE tool, the question remains of if it is implemented will it actually be useable and useful to modelers? • Research can be done to compare ease of use, understanding, and quality of ABPM models in relation to other business process modeling grammars. 112 Bibliography Bosilj-VukSic, V. , & Hlupic V. (2001). Petri Nets and IDEF Diagrams: Applicability and Efficacy for Business Process Modeling. Informatica 25(1), 123-133. CIO Magazine. (1997). Anatomy of a Failure. Retrieved August 29, 2004, from http://www.cio.eom/archive/enterprise/l 11597_data.html Grause, D.C. , & Weinberg, G, M . (1989). Exploring Requirements: Quality Before Design. New York, N Y : Dorset House Publishing. Huckvale, T. & Ould, M . (1994). Process Modeling: Why, What, and How. In K.Spurr, P. Layzell, L. Jennison, & N. Richards (Eds.) Software Assistance for Business Re-engineering (pp. 81-97). West Sussex, England: John Wiley & Sons Ltd. Hui, S. (1997). An Object-Oriented Workflow Management System. M.Sc. Dissertation. Faculty of Commerce and Business Administration, University of British Columbia. I. Jacobson, Object-Oriented Software Engineering, Addison-Wesley, 1992 Introduction to Petri Nets. (2004). Introduction to Petri Nets. Retrieved April 22, 2004 from http://worldserver.oleane.com/adv/elstech/petrinet.htm Jensen, K: A Brief Introduction to Coloured Petri Nets. In: E. Brinksma (ed.): Tools and Algorithms for the Construction and Analysis of Systems. Proceeding of the TACAS'97 Workshop, Enschede, The Netherlands 1997, Lecture Notes in Computer Science Vol. 1217, Springer-Verlag 1997, 203-207 Kemper, P. (2004, February 5). Lecture on Petri-Nets. Retrieved April 22, 2004, from http://www.iai.inf.tu-dresden.de/ms/lvbeschr/vwahl_petri.html Kluge, W.: The Kicking Horse Pass Problem Petri Net News Letters No. 54, (1998), pp. 3-15 ' Knowledge'Based Systems, Inc. (2000, June, 23). IDEF3 Method Report. Retrieved April 22, 2004, from www.idef.com/downloads/pdf/ideO_fn.pdf Tan, W. (William Tan Khoon Lee) (1997). A semantically-enhanced object-oriented CASE tool for enterprise modeling. M.Sc. Dissertation. Faculty of Commerce and Business Administration, University of British Columbia. ' Parsons, J. & Wand, Y: (1997). Using Objects for Systems Analysis. Communications of the ACM, 40(12),-104-110. Petri Nets. (2004). Petri Nets. Retrieved April 22, 2004 from http://www.petrinets.info/ 113 Romney, M . (1994, October). Business Process Re-engineering. The CPA Journal Online. Retrieved August 29, 2004, from http://www.nysscpa.org/cpajournal/old/16373954.htm The Standish Group. (1994) The Chaos Report. Retrieved August 29, 2004, from http://wwwl.standishgroup.eom//sample_research/chaos_1994_l.php Value Based Management.net. (2004) Business Process Reengineering. Retrieved August, 29, 2004, from http://www.valuebasedmanagement.net/methods_bpr.html Wand, Y . & Wang, R. (1996). Anchoring Data Quality Dimensions in Ontological Foundations. Communications of the ACM, 39(1 J) 86-95. Wand, Y . & Weber, R. (1990). Mario Bunge's Ontology as a Formal Foundation for Information Systems Concepts. In Dorn, G. & Weingartner, P. (Eds.), Studies in Bunge's Treatise on Basic Philosophy, the Poznan Studies in the Philosophy of the Sciences and the Humanities (ppl23-150). Rodopi, Amersterdam Wand Y. & Weber, R. (1993). On the ontological expressiveness of information systems analysis and design grammars. Journal of Information Systems, 3, 217-237. Wand, Y. & Weber, R. (1995). On the deep structure of information systems. Information Systems Journal, 5,-203-223. Wand, Y. & Weber, R. (2002). Research Commentary: Information Systems and Conceptual Modeling - A Research Agenda. Information Systems Research, 13(4), 363-376. Wand Y . & Woo. G. (1999): Ontology-Based Rules for Object-Oriented Enterprise Modeling: Working paper. Faculty of Commerce and Business Administration, University of British Columbia. Wand, Y., ! W o o , C , & Jung, D. (2000). Object-Oriented Modeling: From Enterprise to Logical Design. Proceedings of the Tenth Annual Workshop on Information Technologies and Systems (WITS'00, December 9-10, Brisbane Australia), 25-30. Wang, Q. (2002); A'Proposal for a Process Modeling Methodology. M.Sc. Dissertation. Faculty of Commerce and Business Administration, University of British Columbia. Wart, S., Wand, Y. & Woo, C. (1993). Object-Oriented Systems Analysis: An Introduction: Faculty of Commerce and Business Administration, University of British Columbia. Zhang, X. (1998). The Visualization of Object-Oriented Enterprise Modeling. M.Sc. Dissertation. Faculty of Commerce and Business Administration, University of British Columbia. Zhao, H. (1995). Object-Oriented Enterprise Modeling. M.Sc. Dissertation. Faculty of Commerce and Business Administration, University of British Columbia. Zhou, M and Zurawski R: Introduction to Petri Nets in Flexible and Agile Automation. In: M . Zhou(ed.): Petri Nets in Flexible and Agile Automation, Kluwer Academic Publishers, 1999, pp. 1-23 Zimmerman, A . (2004) Petri Nets. Retrieved April 22, 2004, from http://pdv.cs.tu-berlin.de/~azi/petri.html Appendix A - Step By Step Derivation Of The ABPM For The A C M E Warehouse Management Case The purpose of this appendix is to illustrate the development of an A B P M following the A B P M algorithm. We will use the A C M E Warehouse Management Inc. case for this demonstration. The case will be italicized to minimize confusion with the A B P M process At this point we are at step 1 of the algorithm: Identify the external agents A CME Warehouse Management Inc. offers storage facilities and redistribution services (between their different warehouses) across the nation. A customer can request space in a particular warehouse, request items to be transferred to another warehouse, or request withdrawal of items from a particular warehouse (even for items not stored there). For the purpose of this case, we only look at the activities involved in processing a withdrawal request. A customer contacts ACME headquarters to request a withdrawal. The scope of the process has been defined as the activities involved in processing a withdrawal request at the A C M E Warehouse Management facilities. Where do withdrawal requests come from? The customer. Hence the customer must be an external agent. We then move onto step 2 of the algorithm: for each external agent identify the changes generated. Customer generates 2 external changes: a customer contacts ACME headquarters to request a withdrawal, and the customer will come to the warehouse on 116 the required date to pick up the items. Thus, the in terms of our notation changes generated by Customer are 'withdrawal request' and 'arrival' Step 3 of the algorithm: for each change:, means we will look at what happens due to withdrawal request and arrival separately. We will first analyze withdrawal request and then arrival since arrival appears later in the case and logically the customer should not arrive to pick up items until a request for the items has been made. We are at step 3.1 Identify the agent or resource that was changed by withdrawal request A customer contacts ACME headquarters to request a withdrawal. An office clerk checks whether the customer has the authority to withdraw the items. This raises the question of was A C M E headquarters or Office Clerk the agent changed by Customer? The answer is to look at what actually becomes unstable. Does A C M E headquarters or the Office Clerk act next? Thus the answer is Office Clerk. This does raise the possibility however that there could be some sort of composite agent A C M E headquarters of which Office Clerk is a component agent. This possibility will be discussed when we get to step 4. We are at step 3.2. If a resource was changed identify the incoming interface attribute. Office Clerk is an agent since it performs an action. We are at step 3.3 If the agent is an internal agent:, Office clerk is an internal agent since Office Clerk is in the domain of interest. 117 We are at step 3.3.1 If this is the first change to an agent, or the last activity of an agent has gone through a sequence of instability-change-stability create a new activity. This is the first change for Office Clerk so a new activity will be created. We are at step 3.3.2: Identify the incoming interface attributes that were modified. Office Clerk needs some sort of interface attribute to handle the withdrawal request. According to our modeling grammar it will be Customer::Withdrawal Request. We are at step 3.3.3: Identify any state laws that may restrict change. An office clerk checks whether the customer has the authority to withdraw the items. The clerk then passes the withdrawal request to the warehouse where the customer wants to pick up the items and the office clerk will notify the customer as to the status of the request (approved, or declined due to lack of authority, no inventory, or no transportation). And Once the office clerk has recorded the items to be withdrawn, he or she forwards the request to the manager (foreman) of the warehouse. There appear to be no state laws on the incoming interface attributes We are at step 3.3.4 Identify any transformation laws that may exist for the incoming interface attributes. There do not appear to be any state laws restricting further change since the office clerk checks the customer authorization immediately upon receiving the withdrawal request. 118 We are at step 3.3.5: If an agent becomes unstable:, Off ice Clerk does become unstable because it immediately performs an action upon receiving the withdrawal request. We are at step 3.3.5.1: Identify the operations that may occur. An office clerk checks whether the customer has the authority to withdraw the items. The clerk then passes the withdrawal request to the warehouse where the customer wants to pick up the items and the office clerk will notify the customer as to the status of the request (approved, or declined due to lack of authority, no inventory, or no transportation). A n d Once the office clerk has recorded the items to be withdrawn, he or she forwards the request to the manager (foreman) of the warehouse. The warehouse manager is responsible for directing the redistribution of items between warehouses. The Off ice Clerk needs to perform an authorization status check o f that customer and then needs to either contact the appropriate warehouse or notify the customer that they are refused due to not passing the authorization check. In terms o f our modeling grammar notation the services performed are Authority Check and either Contact Warehouse or Noti fy About Status. The clerk recording items is actually part o f the incoming request from Customer 1 4 . We are at step 3.3.5.2 Identify any transformation laws that may affect what operations occur. An office clerk checks whether the customer has the authority to withdraw the items. The clerk then passes the withdrawal request to the warehouse where the customer wants to pick up the items and the office clerk will notify the customer as to the status of 1 4 Since ABPM does not deal with data we do not need to indicate a form being filled out. 119 the request (approved, or declined due to lack of authority, no inventory, or no transportation). And Once the office clerk has recorded the items to be withdrawn, he or she forwards the request to the manager (foreman) of the warehouse. The warehouse manager is responsible for directing the redistribution of items between warehouses. Contact Warehouse and Notify About Status are mutually exclusive15. This means there is a transformation law restricting which operation will happen. Since it is freeform we can have the law read as Contact Warehouse and Notify About Status are mutually exclusive. We are at step 3.3.5.3: Identify the internal attributes that will be affected. An office clerk checks whether the customer has the authority to withdraw the items. The clerk then passes the withdrawal request to the warehouse where the customer wants to pick up the items and the office clerk will notify the customer as to the status of the request (approved, or declined due to lack of authority, no inventory, or no transportation). And Once the office clerk has recorded the items to be withdrawn, he or she forwards the request to the manager (foreman) of the warehouse. The warehouse manager is responsible for directing the redistribution of items between warehouses. The office clerk only uses its own internal information in the authority check operation, the other operations involve interaction with other agents. The notation according to our grammar for this information on customer authorization status will be Authorization Status. 1 5 We assume the office clerk'will not pass a withdrawal request to the warehouse when the customer has not passed an authorization check. 120 We are at step 3.3.5.4: Identify the outgoing interface attributes that were modified. An office clerk checks whether the customer has the authority to withdraw the items. The clerk then passes the withdrawal request to the warehouse where the customer wants to pick up the items and the office clerk will notify the customer as to the status of the request (approved, or declined due to lack of authority, no inventory, or no transportation). And Once the office clerk has recorded the items to be withdrawn, he or she forwards the request to the manager (foreman) of the warehouse. The warehouse manager is responsible for directing the redistribution of items between warehouses. Contact Warehouse modifies an interface attribute that affects another agent; Warehouse Manager. Notify About Status modifies an interface attribute that affects Customer. According to our grammar the notation will be Withdrawal Request::Warehouse Manager and Order Status::Customer We are at step 3.3.4.5: Repeat steps 3.3.4.1 to 3.3.4.4 until the agent becomes stable. Since Office Clerk performs no more actions as a direct result of the incoming external request it can at this point be considered stable. Since we have reached a state where an agent is stable it is useful to show what we have developed. Figure A - l shows where we currently are. Warehouse Manager is purposely left nebulous is this case since we do not know if it is an internal agent or an external agent, or a resource. 121 Customer Office Clerk Activity I Withdrawal Request Affected Attributes Customer::Withdrawal request Authorization Status Withdrawal Request:Warehouse Manager Order St.HiK--rnstnmer Activity 1 Operations Contact Warehouse and Notify About Status are mutually exclusive Authority Check Contact Warehouse Notify about status Warehouse Manager Figure A - l Office Clerk Is Stable We are at step 3.3.6: repeat step 3 for each outgoing interface attribute of an agent that was changed in step 3.3.5.4. The outgoing interface attributes from step 3.3.4.4 were Order Status::Customer and Withdrawal Request::Warehouse. For the sake of conciseness from this point forward we will use a tabular format wherever possible. If the warehouse does not have the items or does not have enough quantity of the items, the warehouse manager will contact other warehouses for the requested items. If the items are located the warehouse manager will ask the planner to arrange for transportation for the requested items and the warehouse manager will notify the office clerk if the request can be fulfilled or not, and the reason. The office clerk will notify the customer as to the status of the request (approved, or declined due to lack of authority, no inventory, or no transportation) and The warehouse manager is responsible for directing the redistribution of items between warehouses. If the items are not all available in the warehouse, transport requests are issued. The warehouse manager fills out a redistribution form with the following information: items to be moved, place from which to take the items, warehouse to transport the items to, quantity to be moved, and the date by when the redistribution must be done. The warehouse manager forwards the form to the planner to organize the interwarehouse transportation of the items 122 Chang e: Order Status::Customer Step Output 3.1 Customer is an existing agent 3.2 Customer is an agent 3.3 Customer is an external agent Table A-l Step 3 for Order Status::Customer Change: Withdrawal Request::Warehouse Manager Step Output 3.1 Warehouse Manager is a new agent 3.2 Warehouse Manager is an agent 3.3 Warehouse Manager is an internal agent 3.3.1 This is a new activity for Warehouse Manager 3.3.2 Office Clerk::Withdrawal Request 3.3.3 No state laws found 3.3.4 No transformation laws found 3.3.5 Warehouse Manager is unstable 3.3.5.1 Check Inventory Contact Other Warehouses Notify About Status 3.3.5.2 Contact Other Warehouse and Notify About Status are mutually exclusive 3.3.5.3 Inventory Information 3.3.5.4 . Item Existence::Warehouse Order Status:":Office Clerk 3.3.5.5 Warehouse Manager is'currently stable Table A-2 Step 3 For Withdrawal Request::Warehouse Manager Since we have reached a state where an agent is stable it is useful to show what we have developed. Figure A-2 shows where we{currently are. Planner is purposely left nebulous is this case since we do not know if it is an internal agent or an external agent. Warehouse is purposely left nebulous is this case since we do not know if it is an internal agent or an external agent 16 From this point forward we will not show the ABPM after each step in order to be more concise. We will however show the full ABPM after all the changes have been addressed. ' Recall that even though many warehouses may be contacted we only need to show one since we are showing agents and not instances of agents in this diagram. 123 Office Clerk Warehouse Manager Withdrawal requesl::Ollice Clerk C Office Clcrk::Ordcr Slatus ' Activity I Withdrawal Request Affected Attributes Customer::Withdra\val request Authorization Slatus Withdrawal Request:Warehouse Manager Order S lnNis -Cnstomcr Activity I Operations Contact Warehouse and Notify About Status are mutually exclusive Authority Check Contact Warehouse Notify about status Activity I Find Items Affected Attributes Office Clcrk::Withdrawal Request Inventory Information Order Slatus::Ollice Clerk l l r m Fv i s l . ' n r r -War i ' ho i i s i -Activity 1 Operations Contact Other Warehouse and Notify About Status are mutually exclusive Check Inventory •Notify about status Contact Other Warehouses Figure A-2 After Step 3 For Order Status::Customer And Withdrawal Request::Warehouse We are at step 3.2.5: repeat step 3 for each outgoing interface attribute of an agent that was changed in step 3.3.5.4. The outgoing interface attributes from step 3.3.5.4 were Item Existence::Warehouse, Transpprt Form::Planner, and Order Status::Office Clerk If the warehouse does not have the items or does not have enough quantity of the items, the warehouse manager will contact other warehouses for the requested items. If the items are located the warehouse manager will ask the planner, to arrange for transportation for the requested items. i C h a n g e : Item Existence::Warehouse ; i 3.1 i Warehouse is a new agent j i 3.2 i Warehouse is an agent ; i :| 3.3. j . Warehouse is an internal agent.:\ ..:;.\..;...v.......ij:..' . i 3.3.1 i This is a new activity for Warehouse I i 3.3.2 i Warehouse Manager::Item Existence j , i. 3.3.3 i No state laws found t -I t i" 3.3.4 I No transformation laws found t < t ! 3.3.5 j Warehouse is unstable i 1 '.' '' "i 3.3.5.1 ' "•; Check Inventory '• • • i I Notify About Search Results j i 3.3.5.2 i No transformation laws found . , i < i : 3.3.5.3 \ Inventory Information I i 3.3.5.4 j Search Results::Warehouse Manager i i 3.3.5.5 i Warehouse is currently stable ' . . . . . . . . . . .J Table A-3 Step 3 For Item Existence::Warehouse 124 We are at step 3.2.5: repeat step 3 for each outgoing interface attribute of an agent that was changed in step 3.3.5.4. The outgoing interface attribute from step 3.3.5.4 was Search Results::Warehouse Manager. If the warehouse does not have the items or does not have enough quantity of the items, the warehouse manager will contact other warehouses for the requested items. If the items are located the warehouse manager will ask the planner to arrange for transportation for the requested items and the warehouse manager will notify the office clerk if the request can be fulfilled or not, and the reason. And If the items are not all available in the warehouse, transport requests are issued. The warehouse manager fills out a redistribution form with the following information: items to be moved, place from which to take the items, warehouse to transport the items to, quantity to be moved, and the date by when the redistribution must be done. The warehouse manager forwards the form to the planner to organize the interwarehouse transportation of the items. Change: Search Results::Warehouse Manager Step Output ' ' 3.1 Warehouse Manager is an existing agent 3.2- Warehouse Manager is an agent 3.3 Warehouse Manager is an internal agent 3.3.1 This is a new activity for Warehouse Manager 3.3.2 Warehouse Manager::Search Results 3.3.3 No state laws found 3.3.4 No transformation laws found 3.3.5 Warehouse Manager is unstable 3.3.5.1 •' Contact Planner 1 . Notify About Status 3.3.5.2 . Contact Planner and Notify About Status are . mutually exclusive 3.3.5.3 No internal attributes found 3.3.5.4 " Transport Form::Planner Order Status::Office Clerk 3.3.5.5 - Warehouse is currently stable A-4 Step 3 For Search Results::Warehouse Manager 125 We are at step 3.2.5: repeat step 3 for each outgoing interface attribute of an agent that was changed in step 3.3.5.4. The outgoing interface attributes from step 3.3.5.4 were Transport Form::Planner and Order Status::Office Clerk. The planner's responsibility is to schedule the company's truck fleet to accommodate requests for transportation, taking into account the existing schedule of each truck and its capacity. The warehouse manager will be notified whether the transportation request can or cannot be satisfied and the planner issues transport orders to truck drivers and The warehouse manager forwards the form to the planner to organize the interwarehouse transportation of the items. The items to be moved are marked as move-pending, and the planner initiates a plan to have the items at the appropriate warehouse at the given date. Once interwarehouse transport plans are finalized, transport requests are issued to the truck drivers. Change: Transport Form::Planner Step Output 3.1 Planner is a new agent .3.2 ' Planner is an agent • • 3.3 Planner is an internal agent 3.3.1 This is a new activity for planner 3.3.2 Warehouse Manager: :Transport Form 3.3.3 No state laws found 3.3.4 No transformation laws found 3.3.5 Planner is unstable • 3.3;5.1 Mark Items Schedule Trucks Notify About Transport , , .. Issue Orders 3.3.5.2 No transformation laws found 3.3.5.3 ~ Inventory Information Truck Information 3.3.5.4. •' Transport Schedule::Warehouse Manager Transport Orders::Truck Driver 3.3.5.5 • Planner is currently stable Table A-5 Step 3 For Transport Form::Planner 126 We are at step 3.2.5: repeat step 3 for each outgoing interface attribute of an agent that was changed in step 3.3.5.4. The outgoing interface attributes from step 3.3.5.4 were Transport Schedule::Warehouse Manager and Transport Orders::Truck Driver. The planner's responsibility is to schedule the company's truck fleet to accommodate requests for transportation, taking into account the existing schedule of each truck and its capacity. The warehouse manager will be notified whether the transportation request can or cannot be satisfied. The warehouse manager will notify the office clerk if the request can be fulfilled or not, and the reason. Change: Transport Schedule::Warehouse Manager Step Output 3.1 Warehouse Manager is an existing agent 3.2 ; • • • Warehouse Manager is an agent 3.3 Warehouse Manager is an internal agent 3.3.1 This is a new activity for Warehouse Manager 3.3.2 Planner: :Transport Schedule 3.3.3 No state laws found 3.3.4 No transformation laws found 3.3.5 Warehouse Manager is unstable 3.3.5.1 Notify About Status 3.3.5.2 No transformation laws found 3.3.5.3,,- , No internal attributes found 3.3.5.4 Order Status::Office Clerk 3.3.5.5 Warehouse Manager is currently stable Table A-6 Step 3 For Transport Schedule::Warehouse Manager We are at step 3.2.5: repeat step 3 for each outgoing interface attribute of an agent that was changed in step 3.3.5.4. The outgoing interface attribute from step 3.3.5.4 was Order Status::Office Clerk. '• . 127 The warehouse manager will notify the office clerk if the request can be fulfilled or not, and the reason. The office clerk will notify the customer as to the status of the request (approved, or declined due to lack of authority, no inventory, or no transportation). Change: Order Status::Office Clerk Step Output 3.1 Office Clerk is an existing agent 3.2 Office Clerk is an agent 3.3 Office Clerk is an internal agent 3.3.1 This is a new activity for Office Clerk 3.3.2 Warehouse: :Order Status 3.3.3 No state laws found 3.3.4 No transformation laws found 3.3.5 Office Clerk is unstable 3.3.5.1 Notify About Status 3.3.5.2 No transformation laws found 3.3.5.3 No internal attributes found 3.3.5.4 Order Status: :Customer 3.3.5.5 Office Clerk is currently stable Table A-7 Step 3 For Order Status::Office Clerk We are at step 3.2.5: repeat step 3 for each outgoing interface attribute of an agent that was changed in step 3.3.5.4. The outgoing interface attribute from step 3.3.5.4 was Order Status::Customer. This change was already dealt with. We can now return to the previous change and repeat step 3 for another of the outgoing interface attributes that was changed in step 3.3.5.4. In this case, webacktrack all the way to the change Transport Form::Planner which has another outgoing change of Transport Orders::Truck Driver. The planner issues transport orders to truck drivers. After receiving a transport order, the truck driver informs the warehouse about the pickup of the items. The warehouse manager will make arrangements to have the items, ready, when the truck arrives. When the truck arrives at the warehouse the items are loaded. The truck driver then informs the next warehouse about the delivery. When the truck has arrived at the next warehouse, the items are unloaded. A warehouse worker finds space for the items and arranges to . - -. • 128 have them moved to the allocated space. The worker updates the warehouse's inventory information. Truck drivers are required to report the status of the truck and the delivery to the planner after each step and The truck driver alerts the warehouse manager of the time he or she will be at the warehouse to pick up the items. The warehouse manager gives appropriate requests to the warehouse worker on the date of delivery to have the items ready for when the truck is expected. When the warehouse worker gets a request to fetch items, he or she, at the appropriate time, orders forklift operators to move the items to the loading platform. The forklift operators execute the internal warehouse operation. When the truck driver arrives, the driver notifies the warehouse worker to have the items loaded into the truck. The truck driver notifies the next warehouse manager when it is expected to arrive at the next warehouse. The number of items in the current warehouse decreases, and the transport request is marked as on transport. Change: Transport Orders::Truck Driver Step Output 3.1 Truck Driver is a new agent '3.2 Truck Driver is an agent' 3.3 Truck Driver is an internal agent 3.3.1 This is a new activity for Truck Driver 3.32 Planner: :Transport Orders 3.3.3 No state laws found 3.3.4 No transformation laws found 3.3.5 Truck Driver is unstable 3.3.5.1 Notify About Pickup - '•>••'.'../,. V '• ' ' Report Truck Status Drive To Pickup Warehouse Notify Ready To Load Report Truck Status 3.3:5.2 No transformation laws found 3.3.5.3 Truck Status 3.3.5.4 •' Truck Status::Planner Pickup Notification::Warehouse Manager Ready To Load::Warehouse Worker, . , Truck Status::Planner 3.3.5.5 Truck Driver is currently stable Table A-8 Step 3 For Transport Orders::Truck Driver 129 We are at step 3.2.5: repeat step 3 for each outgoing interface attribute of an agent that was changed in step 3.3.5.4. The outgoing interface attributes from step 3.3.5.4 were Pickup Notification::Warehouse Manager, Truck Status::Planner, Ready To Load::Warehouse Worker, and Truck Status::Planner. The planner issues transport orders to truck drivers. After receiving a transport order, the truck driver informs the warehouse about the pickup of the items. The warehouse manager will make arrangements to have the items ready when the truck arrives. When the truck arrives at the warehouse the items are loaded and The truck driver alerts the warehouse manager of the time he or she will be at the warehouse to pick up the items. The warehouse manager gives appropriate requests to the warehouse worker on the date of delivery to have the items ready for when the truck is expected. Change: Pickup Notification::Warehouse Manager Step Output 3.1 Warehouse Manager is an existing agent 3.2 Warehouse Manager is an agent 3.3 Warehouse Manager is an internal agent 3.3.1 This is a new activity for Warehouse Manager 3.3.2 Truck Driver::Pickup Notification 3.3.3 No state laws found 3.3.4 No transformation laws found 3.3.5 Warehouse Manager is unstable 3.3.5.1 Notify to Ready Items 3.3.5.2 No transformation laws found 3.3.5.3 • No internal attributes found 3.3.5.4 Ready Items::Warehouse Worker 3.3.5.5 Warehouse Manager is currently stable Table A-9 Step 3 For Pickup Notification::Warehouse Manager We are at step 3.2.5: repeat'step 3 for each outgoing interface attribute of an agent that was changed in step 33.5.4. The outgoing interface attribute from step 3.3.5.4 was Ready Items::Warehouse Worker. - . 130 The warehouse manager gives appropriate requests to the warehouse worker on the date of delivery to have the items ready for when the truck is expected. When the warehouse worker gets a request to fetch items, he or she, at the appropriate time, orders forklift operators to move the items to the loading platform. The forklift operators execute the internal warehouse operation. Change: Ready Items::Warehouse Worker Step Output 3.1 Warehouse Worker is a new agent 3.2 Warehouse Worker is an agent 3.3 Warehouse Worker is an internal agent 3.3.1 This is a new activity for Warehouse Worker 3.3.2 Warehouse Manager::Ready Items 3.3.3 No state laws found 3.3.4 No transformation laws found 3.3.5 Warehouse Worker is unstable 3.3.5.1 Issue Move Item Orders 3.3.5.2 No transformation laws found 3.3.5.3 • No internal attributes found 3.3.5.4 Move Items::Forklift Operator 3.3.5.5 Warehouse Worker is currently stable Table A-10 Step 3 For Ready Items::Warehouse Worker We are at step 3:2.5: repeat step 3 for each outgoing interface attribute of an agent that was changed in step 3.3.5.4. The outgoing interface attribute from step 3.3.5.4 was Move Items: :Forklift Operator Change: Move Items::Forklift Operator Step Output 3.1 Forklift Operator is a new agent 3.2 Forklift Operator is an agent 3.3 Forklift Operator is an internal agent 3.3.1 This is a new activity for Forklift Operator 3.3.2 Warehouse Worker::Move Items' 3.3.3 No state laws found • -3:3.4 No transformation laws found 3.3.5 Forklift Operator is unstable 3.3.5.1 Move Items 3.3.5.2 No transformation laws found 3.3.5.3 Inventory Information 3.3.5.4 No Outgoing Interface Attributes found 3.3.5.5 Forklift Operator is currently stable Table A'-l 1 Step 3 For Move Items::Forklift Operator 131 We are at step 3.2.5: repeat step 3 for each outgoing interface attribute of an agent that was changed in step 3.3.5.4. There are no outgoing interface attributes from step 3.3.5.4 We can now return to the previous change and repeat step 3 for another of the outgoing interface attributes that was changed in step 3.3.5.4. In this case, we backtrack to the change Transport Orders::Truck Driver which has another outgoing change of Truck Status::Planner. The planner issues transport orders to truck drivers. After receiving a transport order, the truck driver informs the warehouse about the pickup of the items. The warehouse manager will make arrangements to have the items ready when the truck arrives. When the truck arrives at the warehouse the items are loaded. The truck driver then informs the next warehouse about the delivery. When the truck has arrived at the next warehouse, the items are unloaded. A warehouse worker finds space for the items and arranges to have them moved to the allocated space. The worker updates the warehouse's inventory information. Truck drivers are required to report the status of the truck and the delivery to the planner after each step and The transportation time is recorded and stored. Change: Truck Status::Planner Step Output 3.1 Planner is an existing agent 3.2 • Planner is an agent • ' 3.3 Planner is an internal agent 3.3.1 • This is a new activity for Planner 3.3.2 Truck Driver: :Truck Status 3.3.3 No state laws found 3.3.4 No transformation laws found 3.3.5 Planner is unstable 3.3.5.1 Record Time . . . . 3.3.5.2 No transformation laws found 3.3.5.3 Transport Information 3.3.5.4 No outgoing interface attributes found 3.3.5.5 Planner is currently stable Table A-12 Step 3 For Truck Status-Planner 132 We are at step 3.2.5: repeat step 3 for each outgoing interface attribute of an agent that was changed in step 3.3.5.4. There are no outgoing interface attributes from step 3.3.5.4 We can now return to the previous change and repeat step 3 for another of the outgoing interface attributes that was changed in step 3.3.5.4. In this case, we backtrack to the change Transport Orders::Truck Driver which has another outgoing change of Ready To Load::Warehouse Worker. After receiving a transport order, the truck driver informs the warehouse about the pickup of the items. The warehouse manager will make arrangements to have the items ready when the truck arrives. When the truck arrives at the warehouse the items are loaded. The truck driver then informs the next warehouse about the delivery and When the truck driver arrives, the driver notifies the warehouse worker to have the items loaded into the truck. The truck driver notifies the next warehouse manager when it is expected to arrive at the next warehouse. The number of items in the current warehouse decreases, and the transport request is marked as on transport. Change: Ready To Load::Warehouse Worker Step Output 3.1 Warehouse Worker is an existing agent 3.2 Warehouse Worker is an agent 3.3 Warehouse Worker is an internal agent 3.3.1 This is a new activity for Warehouse Worker 3.3.2 Truck Driver::Ready To Load 3.3.3 No state laws found '3.3.4 No transformation laws found 3.3.5 Warehouse Worker is unstable 3.3.5.1 Load Truck Mark As On Transport 3.3.5.2 No transformation laws found 3.3.5.3 Inventory Information 3.3.5.4 On Transport: :Truck Driver ' 3.3.5.5 ' Warehouse Worker is currently stable Table A-13 Step 3 For Ready To Load::Warehouse Worker 133 We are at step 3.2.5: repeat step 3 for each outgoing interface attribute of an agent that was changed in step 3.3.5.4. The outgoing interface attribute from step 3.3.5.4 was Loaded: :Truck Driver. When the truck arrives at the warehouse the items are loaded. The truck driver then informs the next warehouse about the delivery. When the truck has arrived at the next warehouse, the items are unloaded. A warehouse worker finds space for the items and arranges to have them moved to the allocated space. The worker updates the warehouse's inventory information. Truck drivers are required to report the status of the truck and the delivery to the planner after each step and The truck driver notifies the next warehouse manager when it is expected to arrive at the next warehouse. The number of items in the current warehouse decreases, and the transport request is marked as on transport. And When the truck has arrived at the next warehouse, the truck driver notifies the warehouse worker to unload the items. The truck driver signs off the job. Change: On Transport: :Truck Driver Step Output 3.1 Truck Driver is an existing agent •3.2. Truck Driver is an agent < • .. 3.3 Truck Driver is an internal agent 3.3.1 This is a new activity for Truck Driver 3.3.2 Warehouse: :On Transport 3.3.3 No state laws found 3.3.4 No transformation laws found 3.3.5 Truck Driver is unstable 3.3.5.1 Notify About Dropoff ' • ' .' - ' Report Truck Status Drive To Dropoff , • , • , •, • Notify Ready To Unload Report Truck Status 3.3.5.2 ' No transformation laws found 3.3.5.3 Truck Status 3.3:5.4' Dropoff Notification::Warehouse Manager Truck Status::Planner Ready To Unload::Warehouse Worker Truck Status::Planner 3.3.5.5 Truck Driver is currently stable Table A-14 Step 3 For On Transport::Truck Driver 134 We are at step 3.2.5: repeat step 3 for each outgoing interface attribute of an agent that was changed in step 3.3.5.4. The outgoing interface attributes from step 3.3.5.4 were Dropoff Notification:: Warehouse Manager, Truck Status::Planner, Ready To Unload::Warehouse Worker, and Truck Status::Planner. The truck driver then informs the next warehouse about the delivery. When the truck has arrived at the next warehouse, the items are unloaded and The truck driver notifies the next warehouse manager when it is expected to arrive at the next warehouse. The number of items in the current warehouse decreases, and the transport request is marked as on transport. Change: Dropoff Notification:: Warehouse Manager Step Output 3.1 Warehouse Manager is an existing agent 3.2 Warehouse Manager is an agent '-• • - - •• • 3.3 Warehouse Manager is an internal agent 3.3.1 This is a new activity for Warehouse Manager 3.3.2 Truck Driver::Dropoff Notification 3.3.3 No state laws found 3:3.4 No transformation laws found 3.3.5 Warehouse Manager is unstable17 3.3.5.1 Notify About Receiving 3.3.5.2 No transformation laws found 3.3.5.3 No internal attributes found 3.3.5.4 Ready Receiving::Warehouse Worker 3.3.5.5 Warehouse Manager is currently stable Table A-15 Step 3 For Dropoff Notification::Warehouse Manager We are at step 3.2:5: repeat step 3 for each outgoing interface attribute of an agent that was changed in step 3.3.5.4. The outgoing interface attribute from step 3.3.5.4 was Ready Receiving::Warehouse Worker 1 7 On the assumption the Warehouse Manager will let the Warehouse Worker know a dropoff is coming, much like how the manager let the worker know a pickup was coming. 135 A warehouse worker finds space for the items and arranges to have them moved to the allocated space. Change: Ready Receiving::Warehouse Worker Step Output 3.1 Warehouse Worker is an existing agent 3.2 Warehouse Worker is an agent 3.3 Warehouse Worker is an internal agent 3.3.1 This is a new activity for Warehouse Worker 3.3.2 Warehouse Manager::Ready Receiving 3.3.3 No state laws found 3.3.4 No transformation laws found 3.3.5 Warehouse Worker is unstable18 3.3.5.1 Find Space Arrange Movement 3.3.5.2 No transformation laws found 3.3.5.3 Warehouse Information Movement Schedule 3.3.5.4 No outgoing interface attributes found 3.3.5.5 Warehouse Worker is currently stable Table A-16 Step 3 For Ready Receiving::Warehouse Worker We are at step 3.2.5: repeat step 3 for each outgoing interface attribute of an agent that was changed in step;3.3.5.4. There are no outgoing interface attributes from step 3.3.5.4 We can now return to the previous change and repeat step.3 for another of the outgoing interface attributes that was changed in step 3.3.5.4. In this case, we backtrack to the change On Transport: :Truck Driver which has another outgoing change of Ready To Unload::Warehouse Worker. When the truck has arrived at the next warehouse, the items are unloaded. A warehouse worker finds space for the items and arranges to have them moved to the allocated space. The worker updates the warehouse's inventory information and When the truck has arrived at the next warehouse, the truck driver notifies the warehouse worker to unload On the. assumption the Warehouse Worker will attempt to have the, warehouse prepared before the truck arrives 136 the items. The truck driver signs off the job. The warehouse workers receive the items and determine a place for them in the warehouse. Forklift operators are told to move the items to the new place in the warehouse. When the truck driver confirms the delivery of the items, the records are updated to reflect the new place for the items. The transportation time is recorded and stored. The redistribution and interwarehouse transport request are marked as performed. The warehouse worker fills in an inventory update form and sends it to the warehouse manager for confirmation and update of the inventory database. Change: Ready To Unload::Warehouse Worker Step Output 3.1 Warehouse Worker is an existing agent 3.2 Warehouse Worker is an agent' - . \ . 3.3 Warehouse Worker is an internal agent 3.3.1 This is a new activity for Warehouse Worker 3.3.2 Truck Driver::Ready To Unload 3.3.3 No state laws found 33.4 " ' • No transformation laws found 3.3.5 Warehouse Worker is unstable 3.3.5.1 Receive Items Determine Place Notify Forklift Operators Ask For Confirmation 3.3.5.2 No transformation laws found 3.3.5.3 Transport Form Warehouse Information 3.3,5.4 Move Items::Forklift Operator Confirmation Needed::Truck Driver 3.3.5.5 Warehouse Worker is currently stable Table A-17 Step 3 For Ready To Unload::Warehouse Worker We are at step 3.2.5: repeat step 3 for each outgoing interface attribute of an agent that was changed in step 3.3.5.4. The outgoing interface attributes from step 3.3.5.4 were Move Items::Forklift Operator and Update Form::Warehouse Manager. Move Items::Forklift Operator has already been dealt with.. 137 The warehouse worker fills in an inventory update form and sends it to the warehouse manager for confirmation and update of the inventory database. Change: Update Form::Warehouse Manager Step Output 3.1 Warehouse Manager is an existing agent 3.2 Warehouse Manager is an agent 3.3 Warehouse Manager is an internal agent 3.3.1 This is a new activity for Warehouse Manager 3.3.2 Warehouse Worker::Update Form 3.3.3 No state laws found 3.3.4 No transformation laws found 3.3.5 Warehouse Manager is unstable 3.3.5.1 Update Inventory Database 3.3.5.2 No transformation laws found 3.3.5.3 Inventory Database 3.3.5.4 No outgoing interface attributes found 3.3.5.5 Warehouse Manager is currently stable Table A-18 Step 3 For Update Form::Warehouse Manager We are at step 3.2.5: repeat step 3 for each outgoing interface attribute of an agent that was changed in step 3.3.5.4. There are no outgoing interface attributes from step 3.3.5.4 We can now return to the previous change and repeat step 3 for another of the outgoing interface attributes that was changed in step 3.3.5.4. At this point we have dealt with all the changes caused directly or indirectly by the external change Withdrawal Request::Office Clerk. We now move onto the changes caused by the external change Arrival "Employee The customer will come to the warehouse on the required date to pick up the items. A warehouse employee will check all the necessary documents and will deliver the items with an accompanying documentation to the customer and When the customer has 138 fetched the items the warehouse workers mark the withdrawal as ready. The items are removed (decreased) from the information system. Change: Arrival "Employee Step Output 3.1 Employee is a new agent 3.2 Employee is an agent 3.3 Employee is an internal agent 3.3.1 This is a new activity for Employee 3.3.2 Customer::Arrival 3.3.3 No state laws found 3.3.4 No transformation laws found 3.3.5 Employee is unstable 3.3.5.1 Check Documents Fetch Items Deliver Items And Documentation Remove Items from the System 3.3.5.2 No transformation laws found 3.3.5.3 Customer Information Inventory Information Inventory Database 3.3.5.4 Items And Documentation::Customer 3.3.5.5 Employee is currently stable . Table A-19 Step 3 For Arrival::Employee We are at step 3.2.5: repeat step 3 for each outgoing interface attribute of an agent that was changed in step 3.3.5.4. The outgoing interface attribute from step 3.3.5.4 was Items And Documentation::Customer. Change: Items And Documentation::Customer Step Output 3.1 Customer is an existing agent 3.2 Customer is an agent 3.3 Customer is an external agent Table A-20 Step 3 For Items And Documentation::Customer At this point we have handled all changes directly and indirectly created due to all external agents. The following tree shows the changes that were generated. Always traversing the tree left shows the order in which the changes were handled. The changes with a box around them are duplicate changes, and the number is used to match the duplicates.. • 139 C u s t o m e r W i t h d r a w a l Request : :O f f ice C l e r k A r r i v a l : : E m p l o y e e Items and D o c u m e n t a t i o n : : C u s t o m e r O rde r Status: :Cus tomer W i t h d r a w a l Request : :Warehouse M a n a g e r Item Ex is tence : :Warehouse I Search Resu l t s : :Warehouse M a n a g e r [Transport F o r m : : P l a n n e r O r d e r S ta tus : :O f f ice C l e r k ITransport F o r m : : P l a n n e r lOrder Status: :O f f i ce C l e r k Transport S c h e d u l e : : W a r e h o u s e M a n a g e r Transport O r d c r s : : T r u c k D r i v e r l Order S t a t u s - O f f i c e C l e r k , Order Status : :Customer P i c k u p N o t i f i c a t i o n : : W a r e h o u s e M a n a g e r I R e a d y I tems: :Warehouse W o r k e r (Truck Status : :P lanner R e a d y T o L o a d : : W a r e h o u s e W o r k e r •I O n T ranspor t : :T ruck D r i v e r T r u c k Status : :P lanner M o v e l tems : :Fo rk l i f t Operator^ D r o p o f f N o t i f i c a t i o n : : Warehouse M a n a g e r I R e a d y R c c e i v i n g : : W a r e h o u s e W o r k e r T r u c k Status : :P lanner R e a d y T o U n l o a d : : W a r e h o u s e W o r k e r T r u c k Status : :P lanner M o v e l tems : :Fo rk l i f t O p e r a t o ^ I Update F o r m : : W a r e h o u s e M a n a g e r Figure A-3 A Tree Showing A l l Of The Changes Figure A-4 shows the solution we have developed at this point We are now at step 4: If needed identify super and subagents using the internal agents. If we stop and consider for a second what it means to be an employee of the warehouse, we can see that it is in fact a super agent with the subagents warehouse worker, warehouse manager, and forklift operator. That is, these agents are all employees that do work at something called a warehouse (More about warehouse will be said in step 5 below). Figure A-5 shows where we currently are. We are now at step 5: If needed identify composite and component agents using the internal agents. We noted above in two places where the possibility of composite agents was suggested. 140 There is no need to create a composite agent for A C M E headquarters. There is only one agent that could be part of such a composition, Office Clerk, and thus no further information would be represented by creating a composite agent. We can create a Warehouse composite agent. The case mentions a warehouse manager, a warehouse worker, and a forklift operator that perform activities at the warehouse, these 141 can become component agents. The attributes and operations of the already existing Warehouse agent can become the emergent attributes and operations of the new Warehouse composite agent. Figure A-6 shows the composite agent warehouse in the decomposed view. When we collapse the decomposed warehouse agent into its aggregate form we get figure A-7. We have now gone through all the steps and are finished. 142 us I Wihdmil Requcu J Aleged Aimhum •net. :W.ihdi.l lequcil | Oder Mum «fli:« clci A(levied Anbmei TrwV Otitr.Pitlup NMIKIMIIT ADMcd An hum I K u i M l Activity 5 Ni»ify ihoil *m»(T Tn«*Ilfim.:[>.i|).»f N.KilkMNin * R«ls Recti v«if :Wirch»iv Wider Vti.> J npetakW k>Rc*d> R«.n»nt -ACUMI) fr I'ptuilt fcvcnK*) t>mhic A %acd AnnhutEi Winhouic W»icr.I'pdwc Fin"— A fleeted Arm bum lruck Itnva.Rad) Toln •»H>« Schedule iWirehiw Actiuv 1 Head} Racitne Altnicd Annhuin Wa-dmiw Slim per: Rent) Ri Avium 11 nl.«J Truil. Alnled \tlnhtiei liulkiw. Rnd> Tit'nli I runnel li«m MiocHeim.lilnOpnoKi tc W.rto.AW ken A Held An hum Pickup ".iSuiw ;Wneruiw Muuyti I ArViKd Anhum DrifnfT S.ini.aHin'>itieN«>w Manes | Re*d> Tnl ni—) :t>ch.< Acimh. : DlwHuia. Si.K Ahu Ihin'T -tR Rad> r.il'nl.*d 4Rep.i Irwi Siui Figure A-5 Creating An Employee Superagent 143 •PS L'mtumcr: Withdrawal rcquei Withdrawal ReuueirWarehi.u'. j jr Jci. .Smut:. C um m i Acimt. 2 Nmif. annul imlcr (*l» Afected Atribute* t hdd Nlatin: Cmtim'i Ai.-n.ns J 1 ipcr>ti.«» •N.«ir> ihiwi uatui [heel Inwnlun N«ils abcui iLMm Cunisct * *her Warehouvei Annus : Decide if order cu A(Tted Atribute! Notify atxiul itatm Acism 1 Notify about order Aftecied At ti i hum Planner: I rampurt Schedule* Arti.> J Operation. Aloted Atributes. Truck I )mer.: Pickup Noulicatior Kead> Items :W«ch.K»c Worker Truck l)nver.:I>ip.n"Nitl.'kilm— Heads Recing "Witrehnuw Worker AciMt. .sfVca™, N. ft UpdJtc ln.enlir> IMahne A fTctld An hulvi Warehouse Worker.:! Jpdme Fornr> Acti.> I Prepare fin pick Ali ted AintHim Warehouse MaiutEf :Keac AITcld Altibt I nut l)river:R Im colors Infori I Activity I Arange Transport Afected Alnhulei i Warehouse Manager: I ran^nofl pom n senior. In fm manor I Truck aifc.mu.on Tranifmrt Schedule ; Warchouvr Manager | 1ran\ra>rt nrdeti:Truck IVt.r .it> l rVmt»™ ACIISIIS 4 Unload Truck Afected Alnhulei Truck Ifciwr::Rc«i) To Unload •— | Warehouie biformtion Mow Items .FortJift I Iperjtof Transput Form | I Ipdae Form .Warehouse Manager Detrmine Place Notify ForklitH>peritoti MarV Trampon Reqursi AJ Perform. -Send Update form Act|.i)_) Search Fur Item) Alecied Annhulcj * WarthouK MmJisuHcm liviletKt biAcnitu] InMnulion Search Rt suits .Warehnuw Manajrt Acti«11>. 1 .L^perot wfl* Check taiiemi*) Nmib about x'lrch rnuhi .'archou* Worker: :M.s* It. Truck Dmc iraospuel In | AITected Altnlxiei ner.;~Iransp.iri order* atehnuw Manager Truck Sun» lljnnrt | Alice led Annhuie rhouse Worker:., l>upif Truck Suiut | Truck Statin Is tolnload ;Wareh.»eW k SLilus Planner •f Nuih Annul Dropof • Report fruck Slum e Tol)ri.>.fT Figure A-6 The Decomposition of Warehouse 144 Offlc* Clerk in Withdrawal request ::OlTicc Ollice Clerk:;Ordcr Status • iviiy 1 Withdrawal Request cd Attributes ithdrawul request A u l hi ui/alii HI Status Withdrawal Request: Warehouse Order SlatuvTusinmcr 4A Heele Customer:^' Activity I Operations Contact Warehouse and Notify Ahou Status are mutually exclusive Authority Check Contact Warehouse-Noiifv annul slulus Aciivii)' 2 Noiily annul order statu? All'eetcd Altribuies Warehouse:;! >rdcr S t a t u s • " Order Status: :Cusiomer 3 Opera Noli IV about Mai i vat:: Ware house ise::1 terns and IXtcumeniation1 Aciivii) I Kind llerm A (levied Attributes Ol lke Clerk-Withdrawal Request Inventor) Informal km Order Status::! Wive Clerk Mem l-xiMemy-WaivhmiM-Activity 1 Operatkms Contact Other Warehouse and Notify About Status are mutual I) exclusive Cheek Inveniory Noiily ahivui status Contati Other Warehouses Aetivit) 2 Search For Items All'evted Attributes Warehouse:: Item IMsiencv lmcnitir> Information Search Results:: Warehouse Avlivilv 2 Operations Check Inventory Noiifv ahoul seiireh results Aeiivitv 3 [Wide if order va Articled Aiiribuics •Warehouse::Search results Transport Tnnn:: Planner Onk-r SI:IIIIS"OIIK-L- cirri Aeliviiy 3 Operations Contact planner and noiily ahoul slulus are mutually exclusive Activity 4 Notify about order Mulus Allevied Aiiribuics Planner:: Transport Schedule « Order Slums Activity 1 Operaiion* Notify about status Aciivity i Prepare for pickup All'evted Attribuies Truck Driver:: Pick up Not i Ileal i Aetivit) 1 Opcrulkms Move (term Aciivii) h Ixvud I ruck All'evted Attributes I ruck DrivcrRead) To Uiod • Inventory Inliirmolion )n Transport "Truck Driver Activity 6 Opera Ijiad Truvk Mark As On Transport -Aciivity 7 Ready Receiving Affected Attribuies Truck Drtvcr;;l)nipulTNotifii Warehouse Informal kin Movement Schedule "Aciivify 7 Operations t'ind Space Arrange Movement Aeiivitv K Unload I ruck Allevied Aiiribuics Tniek Driver::Reudy To Unload * Transport Form Warehouse Information Inventory Information 1 ransport Form Inventory Dalabasc Activity H UpcralKins Receive Items Delermine I'liicc Move Items Mark Transport Request As Performed llpdule Inveniory DtiUihasc Aetiv i|v V Kufil Pickup A (levied Ailrihute* ;:ArrivaJ Customer In formal kin Inveniory Information Items And Document at k> Inventors Database Aviiv try v Operation* Check Documents Fetch Items 4)elivcr Items And Documcnuitxm Remove Items from the Sysli Aeiivitv I Arrange Transport Allevied Aiiribuics Warehouse:Transport Form Inventory In formal km Truvk In format iun Transport Schedule::Warehouse Transport ordcrs::Truck Driver Activity I Operations Mark Inventory Schedule Trucks Notify About transport Issue Orders Activity 2 Record Time AlTeeied Attributes Truvk Driver Truck Status Transport In lor Activity 2 Operai Record Time Activity I Proceed lo pickup AlTeeied Attributes • Planner:: Transport orders Pickup Nolificalion::Warehim.se | Truck Slat us:: Planner Truck Status Read) lo loud:: Wore house Truck Status::Planner Aclivirv I Opcralkxis -Noiifv about pickup • Report truck slulus Drive lo pickup u are house -Notify read) lo kvad • Report truvk status Aciivity 2 Proceed lo dropolT j AtTcctcd Attribuies A' are house:: On Transport Dn Existence:: Warehouse Contact Other Warehouses and Notify About Status are mutually exclusive Contact Other Warehouses Order Status::Office Clerk Notify about status 2 Warehouse:: Item Existence Inventory Information '• Check Inventory Search . Results:: Warehouse .: — v ' . l '< Notify about search results 3 • . • • ' . : : Warehouse:: Search results . ; ' ! ! ' • :• Transport 1 ' Form::Planner-Contact planner and notify about status are mutually exclusive Contact planner Order Status: :office .. clerk Notify about status 4 ' ' ' Planner: transport 149 Schedule Order Status: :office clerk Notify about status 5 Truck Driver::Pickup Notification Inventory Information Move Items 6 Truck Driver::Ready To Load Inventory Information Load Truck On Transport: :Truck Driver Mark As On Transport 7 Truck Driver::Dropoff Notification Warehouse Information Find Space Movement Schedule Arrange Movement 8 Truck Driver::Ready To Unload Transport Form Receive Items Warehouse Information Determine Place Inventory Information Move Items Transport Form Mark Transport Request As Performed Inventory Database -Update Inventory Database 9 Customer::Arrival Customer Information Check Documents Inventory Information Fetch Items Items And Documentation: :Customer Deliver Items And Documentation Inventory Database Remove Items from the System Figure B-9 Warehouse Agent Template in 'composited' view 150 Warehouse Worker Superagent: Employee Attributes Operations Activity State Law Interface Attributes Internal Attributes Transformation Law Operation 1 W a r e h o u s e M a n a g e r : : R e a d y I t e m s M o v e I t e m s : : F o r k l i f t O p e r a t o r I s s u e M o v e I t e m O r d e r s 2 T r u c k D r i v e r : : R e a d y T o L o a d I n v e n t o r y I n f o r m a t i o n L o a d T r u c k O n T r a n s p o r t : : T r u c k D r i v e r M a r k A s O n T r a n s p o r t 3 W a r e h o u s e M a n a g e r : : R e a d y R e c e i v i n g W a r e h o u s e I n f o r m a t i o n F i n d S p a c e M o v e m e n t S c h e d u l e A r r a n g e M o v e m e n t 4 T r u c k D r i v e r : : R e a d y T o U n l o a d T r a n s p o r t F o r m R e c e i v e I t e m s W a r e h o u s e I n f o r m a t i o n D e t e r m i n e P l a c e M o v e I t e m s : : F o r k l i f t O p e r a t o r N o t i f y F o r k l i f t O p e r a t o r s •• T r a n s p o r t F o r m M a r k T r a n s p o r t R e q u e s t A s P e r f o r m e d U p d a t e F o r m : : W a r e h o u s e M a n a g e r S e n d U p d a t e F o r m Figure B-10 Warehouse Worker Agent Template 151 Warehouse Manager Superagent: Employee Attributes Operations Activity State Law Interface Attributes Internal Attributes Transformation Law Operation 1 Office Clerk::Withdrawal Request Inventory Information Check Inventory Item Existence: :Warehouse Contact Other Warehouses and Notify About Status are mutually exclusive Contact Other Warehouses Order Status::Office Clerk Notify about status 2 Warehouse::Search results Transport Form::Planner Contact planner and notify about status are mutually exclusive Contact planner Order Status: :office clerk Notify about status 3 Planner: :Transport Schedule Order Status::office clerk Notify about status 4 Truck Driver::Pickup Notification Ready I terns:: Warehouse Worker - Notify to Ready Items 5 Truck Driver::Dropoff Notification Ready Receiving::Warehouse Worker Notify to Ready Receiving 6 Warehouse Worker::Update Form . „ • . Inventory Database Update Inventory Database Figure B-ll Warehouse Manager Agent Template 152 Appendix C - A Discussion of the BWW Ontology The following discussion of the BWW Ontology is based on Parsons and Wand (Parsons and Wand, 1997), Wand & Weber (Wand and Weber, 1993, 1995), and Wand and Woo (Wand and Woo, 2002). Ontological constructs are italicized as they are introduced. C.l Static Model of things The real world is comprised of things. There are simple things. A composite thing is made up of other things (be it other composite things or simple things or some combination thereof). Things possess properties. Properties are either intrinsic (possessed solely by one thing) or mutual (shared with one or more other things). Properties exist independent of people being aware of their existence. Attributes are not necessarily possessed by a thing, but may be assigned to things by people in order to measure the property. For example, a property of a thing could be the ability to reflect a wavelength of light. People will then attribute a color to this thing. Composite things possess hereditary properties that belong to its component things. Composite things also possess emergent properties that are not possessed by any component thing. As an example a car is a composite thing. It has the hereditary property burns gasoline, from one of its component things the gasoline injected combustion engine. It also possesses the emergent property maximum acceleration which is not possessed by any one component thing. 153 The set of values for the attributes of a thing comprise the state of the thing. A conceivable state space for a thing is the set of all possible states a thing may ever assume. State laws serve to restrict the values of the properties of a thing to a subset of the conceivable state space. State laws must enforce a restriction due to either natural or human laws. A law is a property. For example, most bank accounts have the restriction (state law) that the balance must be greater than or equal to zero. This is due to the human law that people are only allowed to spend up to the total amount of money that is in their bank account. The lawful state space of a thing is the set of states that exist for a thing that comply with its state laws. A lawful state space is usually a subset of the conceivable state space. A class is a set of things that all possess a common property. A kind is a set of things that possess two or more common properties. It should be noted that common property does not refer to mutual property. All things that are part of a natural kind will possess the same lawful state space. C.2 Dynamic Model of things Ontology has the principle that every thing changes and that every change is the change of properties of things. This is an event. Since the measures of properties of a thing comprise the state of the thing, and an event is changing properties. An event is really the change of the state of a thing. Ontology follows the principle of nominal invariance. That is, a thing can change and still be the same thing. An event is carried out via 154 transformation (defined below). The event space is the set of all possible events that can occur in a thing. A transformation is a state (attribute) change from one state to another state. A lawful transformation defines the events that are lawful for a thing. The lawful event space is usually a subset of the event space, and defines those events in a thing that are lawful. The history of a thing is the chronologically ordered states that it has traversed. C.3 Static Model of systems A coupling occurs when one thing acts on another thing that is, the existence of one thing affects the history of another thing. When this situation arises among two things they are said to be coupled (or interact). A set of things can be called a system if when the set of things is bipartitioned there are couplings among the things in the two subsets. The things that make up the system are its system composition. A system environment is composed of the things that are not in the system but interact with things in the system. Systems have a system structure. A system structure consists of the couplings among the things in the system, and the couplings among things in the system and things in the system environment. There exist subsystems. A subsystem is a system of which the composition and structure are subsets of the composition and structure of another system. A system may be broken down into a system decomposition. A system decomposition is a set of subsystems in 155 which every component of the system is either one of the subsystems, or included in the composition of one of the subsystems in the decomposition. Generally, system decompositions require some sort of level structure. A level structure is a partial ordering over the subsystems in the decomposition. The level structure defines which subsystems are components of the system or other subsystems. C.4 Dynamic Model of systems Things, subsystems, and systems do not begin to change unless they are given impetus from an external event (defined below). The state a thing, subsystem, or system remains in unless it is forced to change by an external event is called a stable state. An unstable state is a state that must be changed to another state by actions within the thing, subsystem, or system. Unstability occurs due to the transformation laws that exist within a thing, subsystem, or system. The end result of the state change caused due to instability can be either a new unstable state or a stable state. An external event is an event that affects a thing, subsystem, or system caused by some other thing in the environment of the thing, subsystem, or system. Before an external event occurs the thing, subsystem, or system is in a stable state. After an external event the thing, subsystem, or system may be stable or unstable. An internal event occurs in a thing, subsystem, or system due to lawful transformations of the thing, subsystem, or system. Before an internal event the thing, subsystem, or system is in an unstable state. After an internal event the thing, subsystem, or system may be in either a stable state or an unstable state. 156 Besides being internal or external, events can also be well-defined or poorly defined. A well-defined event is one in which the state of the thing, subsystem, or system after the event can be predicted just by knowing the state of the thing, subsystem, or system prior to the event. A poorly-defined event is one in which the state of the thing, subsystem, or system after the event cannot be predicted just by knowing the state of the thing, subsystem, or system prior to the event. 157 Appendix D - Modeling Grammar Examples The case used in our examples is the same one used in chapter three and is presented again here in its entirety. The A C M E Warehouse Management Inc. Case A C M E Warehouse Management Inc. offers storage facilities and redistribution services (between their different warehouses) across the nation. A customer can request space in a particular warehouse, request items to be transferred to another warehouse, or request withdrawal of items from a particular warehouse (even for items not stored there). For the purpose of this case, we only look at the activities involved in processing a withdrawal request. A customer contacts A C M E headquarters to request a withdrawal. An office clerk checks whether the customer has the authority to withdraw the items. The clerk then passes the withdrawal request to the warehouse where the customer wants to pick up the items. If the warehouse does not have the items or does not have enough quantity of the items, the warehouse manager will contact other warehouses for the requested items. If the items are located the warehouse manager will ask the planner to arrange for transportation for the requested items. The planner's responsibility is to schedule the company's truck fleet to accommodate requests for transportation, taking into account the existing schedule of each truck and its capacity. The warehouse manager will be notified whether the transportation request can or cannot be satisfied. The warehouse manager will, notify theoffice clerk if the request can be fulfilled or not, and the reason/The office clerk will notify the customer as to the status of the request (approved,^ or declined due to lack of authority; no inventory, or no transportation). The planner issues transport orders to truck drivers. After receiving a transport order, the truck driver informs the warehouse about the pickup of the items. The warehouse manager willmake arrangements to have the items ready'when the truck arrives. When the truck arrives at the warehouse the items are loaded. The'truck driver then informs the next warehouse about the delivery. When the truck has arrived at the next warehouse, the items are unloaded. A warehouse worker finds space for the items and arranges to have them moved to the allocated space. The worker updates the warehouse's inventory information. Truck drivers are required to report the status of the truck and the delivery to the planner after each step. -Based on a case in I. Jacobson, Object-Oriented Software Engineering, Addison-Wesley; 1992 158 The customer will come to the warehouse on the required date to pick up the items. A warehouse employee will check all the necessary documents and will deliver the items with an accompanying documentation to the customer. Supplemental description Once the office clerk has recorded the items to be withdrawn, he or she forwards the request to the manager (foreman) of the warehouse. The warehouse manager is responsible for directing the redistribution of items between warehouses. If the items are not all available in the warehouse, transport requests are issued. The warehouse manager fills out a redistribution form with the following information: items to be moved, place from which to take the items, warehouse to transport the items to, quantity to be moved, and the date by when the redistribution must be done. The warehouse manager forwards the form to the planner to organize the interwarehouse transportation of the items. The items to be moved are marked as move-pending, and the planner initiates a plan to have the items at the appropriate warehouse at the given date. Once interwarehouse transport plans are finalized, transport requests are issued to the truck drivers. The truck driver alerts the warehouse manager of the time he or she will be at the warehouse to pick up the items. The warehouse manager gives appropriate requests to the warehouse worker on the date of delivery to have the items ready for when the truck is expected. When the warehouse worker gets a request to fetch' items, he or she, at the appropriate time, orders forklift operators to move the items to the loading platform. The forklift operators execute the internal warehouse operation. When the truck driver arrives, the driver notifies the warehouse worker to have the items loaded into the truck. The truck driver notifies'the next warehouse manager when it is expected to arrive at the next warehouse. The number of items in the current warehouse decreases, and the transport request is marked as on transport. ' '- ' ' When the truck has arrivedat the next warehouse, the truck driver notifies the warehouse worker to unload the items. The truck driver signs off the job. The warehouse workers receive the items and determine a place for them in the warehouse. Forklift operators are told to move the items to the new place in the warehouse. When the truck driver confirms the delivery of the items, the records are updated to reflect the new place for the items. The transportation time is recorded and stored. The redistribution and interwarehouse transport request are marked as performed; The warehouse' worker fills in an inventory update form'and'sends it to the warehouse manager for confirmation and update of the inventory 'database. - ... : When the customer has fetched the items the warehouse workers mark the withdrawal as ready. The items are removed (decreased) from the information system.' -D.l Colored Petri Nets Figure D-l !shows the Colored1 Petri Net solution to the A C M E Case!- 1 1 159 Fetch Items Check Documents Pickup Order d - approved (a,b, c) Remove Items From The System Deliver Items And Documentation ,(a.b) ( Withdrawal (a, b, Request Authority Check Receive ; Find Items Space q / Arrange Movement Move Items Issue Move Item Orders Mark Transport 1 Request As Performed Check For Items d != approved Search Results else 1'd = lack of inventory Drive To Dropoff Warehouse Arrange Transport Drive To Pickup Warehouse Report Truck Status Record Time Mark h Inventory Schedule Trucks Figure D-l Colored Petri Net for the ACME Warehouse Case Declarations: Token String Array var h: Inventory Marked var a, g: Warehouse var i : Trucks Scheduled var b: Items var j : Notified About Transport var c: Customer var k: Truck Schedule var d: Status var 1: Ready To Report var e: Items Not Found var m: Notified About Pickup var f: Items Found var n: Report Done var o: At Pickup var p: Loaded var q: Items Need To Be Moved var r: Movement Orders Issued var s: Notified About Dropoff var t: At Dropoff var u: Space Found var v: Movement Arranged var w: Items Received var x: Place Determined var y: Transport Request Marked var z: Update Form var A: Customer Arrived var B: Documents Checked var C: Items And Documentation Delivered Withdrawal Request > Notify About Status; Table D-l Declarations for the ACME Warehouse Colored Petri Net In a Colored Petri Net the ellipses are called places and represent locations or resource stores. They represent the input and output of transitions. The rectangles are called transitions. Transitions are events, activities, or changes of state. Transitions create or destroy tokens. The arrows are directed arcs and represent necessary pre and post conditions of transitions. Each place has markers called tokens that contain a data value. The declarations table tells the,type of value each token can takes. A 'token' declaration would be just there whereas a 'string array' declaration can hold a list of data. The declaration table also tells which transitions have priority when two or more of them are enabled at the same time. -161 D.2 Integrated Definition 3 Figure D-2 contains the process flow description for the A C M E case. Figure D-3 contains the object state transition network for the A C M E case. In the process flow description the number box is a unit of behaviour (UOB). A UOB is a situation that happens. The arrows represent precedence links, with what happens at the tail of the arrow occurring before what happens at the head of the arrow. An arrow with a triangle on it means that the item at the tail of the arrow must precede the item at the head of the arrow. An arrow with a star means an the item at the head of the arrow must come after the item at the tail of the arrow and the item at the tail of the arrow must precede the item at the head of the arrow. The boxes with letters in them denote junctions. Junctions can be either and (&), or (O), or exclusive or(X), one vertical line in a junction denotes it is asynchronous, two vertical lines note it is synchronous. A UOB with either one or two vertical lines around the name is not a UOB but a referent. One vertical line denotes it is a call and continue referent, two vertical lines denote it is a call and continue referent. An Object State Transition Network describes the states an object travels through. The circles represent object states. The arrows are the links. The boxes are referents to UOBs in the process flow description. The circles with letters in them are junctions of either and (&), or (O), or exclusive or (X). Referents connected at the same point begin at the same time. Referents connected to a circle means the order in which the occur is unknown but they can begin to occur then, and happen before other referents further down the line. 163 UOB/Send Update Form UOB'Move Items UOB/Determirn Place UOB/Receive Items UOR'Notify Ready To Unload UOB'Drive To Dropoff UOB/Find Space UOB/Notify To Ready Receiving UOB/Record Time UOB/Rcport Truck Status I UOB Notify Ready To 1 Load 35 33 31 30 29 27 26 25 37 16 20 UOB/Updaie Inventory Database UOElMaA Transport Request As Performed 36 34 UOB/Record Time UOB/lssue Move Item Orders UOB'Make Pickup Arrangements •v ^ fr-UOB/bsue Orders UOB'Notify •Forklift Operators UOB/Record Time UOB/Rcport Truck Status UOB/Anange Movement UOB'Record Time UOB/Report Truck Status UOB/Notify About Dropoff UOB-Mari As On Transport UOB/Load Truck UOB/Move Items UOB/Dme to Pickup Warehouse UOB/Report Truck Status UOB/Notify About Pickup 32 37 16 28 37 16 24 23 22 21 18 16 15 Figure D-3 Object State Transition Network for the ACME Warehouse Case 164 D.3 Event-controlled Process Chains Figure D-4 contains the EPC for the A C M E case. In EPC a rounded rectangle overlapping a hexagon denotes a process. A hexagon denotes an event, when something happens. A rounded rectangle denotes a task, what happens. An oval with a dashed vertical line denotes an organizational unit, who does the task. A rectangle denotes an information object, data accessed, created, or changed. A dashed arrow denotes a control flow. Control flows connect events to tasks, and tasks to events. A solid arrow denotes an information/material flow, and connects information objects to tasks. A line denotes an organization assignment, and connects organizational units to task. Events and tasks may be combined using circles with either an inverted v, a v, or X O R in them. These correspond to and, or, and exclusive or. Vithdrawal Request, Notify About Pickup Truck J Driver At Dropoff^ Notify Ready To Unload Warehouse Find Space Information Drive To Truck Dropoff Status Space Found Movement • Arrange Schedule Movement /MovemenN. XArranged / • 0 © ~ 0 0 Warehouse^ J Worker^