UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Knowledge identification phase of natural language analysis Browse, Roger Alexander 1977

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

Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.

Item Metadata

Download

Media
831-UBC_1977_A6_7 B76.pdf [ 4.24MB ]
Metadata
JSON: 831-1.0051787.json
JSON-LD: 831-1.0051787-ld.json
RDF/XML (Pretty): 831-1.0051787-rdf.xml
RDF/JSON: 831-1.0051787-rdf.json
Turtle: 831-1.0051787-turtle.txt
N-Triples: 831-1.0051787-rdf-ntriples.txt
Original Record: 831-1.0051787-source.json
Full Text
831-1.0051787-fulltext.txt
Citation
831-1.0051787.ris

Full Text

A KNOWLEDGE IDENTIFICATION PHASE OF NATURAL LANGDAGE ANALYSIS by Roger Alexander Browse B . S c , M c G i l l U n i v e r s i t y 1972 THESIS SUBMITTED IN; PARTIAL FULFILLMENT THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE i n the Department of Computer Science We accept t h i s t h e s i s as conforming to the r e q u i r e d standard The U n i v e r s i t y o f B r i t i s h Columbia January 1977 Q) Ro-ger Alexander Browse, 1977 In presenting th i s thes is in pa r t i a l fu l f i lment of the requirements for an advanced degree at the Univers i ty of B r i t i s h Columbia, I agree that the L ibrary sha l l make it f ree ly ava i l ab le for reference and study. I fur ther agree that permission for extensive copying of th is thesis for scho lar ly purposes may be granted by the Head of my Department or by his representat ives. It is understood that copying or pub l i ca t ion of th is thes is for f inanc ia l gain sha l l not be allowed without my writ ten permission. Department of Computer S c i e n c e The Univers i ty of B r i t i s h Columbia 2075 Wesbrook Place Vancouver, Canada V6T 1W5 Feb. 2 1977 e i a b s t r a c t Case o r g a n i z a t i o n of verbs has provided a powerful mechanism f o r n a t u r a l language a n a l y s i s systems. However, o n l y simple semantic-marker-like i n f o r m a t i o n has been used to determine the a c c e p t i b i l i t y of l e x i c a l elements as c a s e - r o l e f i l l e r s . A c t u a l l y , t h i s a b i l i t y i s i n f l u e n c e d by more i n t r i c a t e r e l a t i o n s among words. In a d d i t i o n , a case-bassd view of semantic knowledge o f t e n leads to the separate s p e c i f i c a t i o n of each shade of meaning of a verb. These two problems are addressed i n t h i s t h e s i s . A c a s e - l i k e o r g a n i z a t i o n o f semantic knowledge which i n c l u d e s a network of r e l a t i o n s among l e x i c a l elements i s presented. Any p i e c e of i n f o r m a t i o n contained i n the system may be used as a case-frame s p e c i f i c a t i o n , or i t co u l d be used as i n f o r m a t i o n which determines c a s e - r o l e f u l f i l l m e n t . Rules f o r the use of t h i s i n f o r m a t i o n have been designed to permit a s i n g l e case-frame to re c o g n i z e many shades of meaning of a verb, even t o the p o i n t of a c c e p t i n g metaphoric language use, The network of r e l a t i o n s i s h i e r a r c h i c a l l y o r g a n i z e d , and knowledge i s r e t a i n e d at many l e v e l s of g e n e r a l i z a t i o n . Along with the e x i s t e n c e of c a s e - o r g a n i z a t i o n i n the network, these m u l t i p l e l e v e l s provide some c o n t r o l over the t r a v e r s a l of the network. i i A s m a l l implementation i s provided to demonstrate the use of a v a r i e t y of s t r a t e g i e s f o r f i t t i n g case-frames to i n p u t . The model i s intended as a bottom-up component f o r the i d e n t i f i c a t i o n of those pieces of i n f o r m a t i o n which may be r e l e v a n t to a given i n p u t . T a b l e of Contents i i i Table of Contents 1. Semantic P r o d u c t i v i t y 1 1a. F o l l o w i n g the Syntax Paradigm 2 1b. Extending the Syntax Paradigm 4 1c. The Impact of Case <...<.... .•<.».««......<.....>••«•« . 5 1d. The S p e c i f i c a t i o n of Semantic A t t r i b u t e s .............. 7 1e, W i l k s 1 Approach 9 1f. What Has Been Ignored 11 2. O b j e c t i v e s and Overview .................................. 14 2a. B r i e f O u t l i n e .....14 2b. Some S p e c i f i c O b j e c t i v e s . 1 7 3. Some E x p l a n a t i o n s 18-3a. A Two-Process View of Comprehension 18 3b. Comprehension Without Representation ................. 21 4. A Knowledge I d e n t i f i c a t i o n System ...25 4a. The Basic Version 25 4b. H i e r a r c h i e s And G e n e r a l i z a t i o n s ...................... 28 4c. Verb H i e r a r c h i e s and More G e n e r a l i z a t i o n 34 4d. Forming S u b - P r o p o s i t i o n s 39 4e. Expanding the Content of the Prototype ............... 42 5. Some Fu r t h e r Issues 48 5a. T r a c i n g 4 8 5b. D i r e c t e d I n s t a n t i a t i o n ..... .... ........ .... .... .... .. 52 5c. Metaphor 54 5d. Le a r n i n g 57 6. An Implemented Model . .... .... .. ... ... .. .... .... .... .... .. 59 T a b l e of contents i v 5a. Sample Huns 60 6b. D i s c u s s i o n .. ......... ..... .. ............ 68 7. Summary and C o n c l u s i o n s .................................. 72 8. Program L i s t i n g s ......................................... 74 9. B i b l i o g r a p h y ....................................... .... .. 79 Semantic P r o d u c t i v i t y 1 1. Semantic P r o d u c t i v i t y The semantic component of a n a t u r a l language a n a l y s i s system must provide some method f o r the s p e c i f i c a t i o n -of those u t t e r a n c e s which are meaningful to the system. E v a l u a t i o n of t h i s component i s u s u a l l y made with with r e s p e c t t o : 1) The o v e r a l l performance of the system. 2) The ease with which counter-examples can be d e v i s e d . Semantic productivity i s an i n t u i t i v e measure of how well the method of s p e c i f i c a t i o n covers a wide range of meaningful p o s s i b i l i t i e s , without i n c l u d i n g them a l l e x p l i c i t l y . , Semantic p r o d u c t i v i t y i s more than j u s t a measure of the degree to which the word-use knowledge i s i m p l i c i t i n the system. The n a t u r a l use of language i s c r e a t i v e , and p a r t of t h a t c r e a t i v i t y i s d e r i v e d from being a b l e t o a l t e r shades of meaning, and even c r e a t e metaphors (see Bobrow, 1975:9; C o l l i n s and Q u i l l i a n , 1972:346). The a b i l i t y t o understand many shades o f meaning can be obtained by: 1) d e f i n i n g each i n s t a n c e d i r e c t l y , and " l y i n g i n wait" f o r each v a r i a n t . 2) Allowing a s i n g l e d e f i n i t i o n to be used i n a f l e x i b l e enough f a s h i o n t o permit v a r i a t i o n s . One argument i n favour of the l a t t e r i s t h a t the s e t of Semantic P r o d u c t i v i t y Semantic P r o d u c t i v i t y 2 p e r m i s s i b l e v a r i a t i o n s i s u s u a l l y determined by the nature of the r e s t of the knowledge i n the system. How w e l l v a r i a t i o n s i n meaning have been l o c a l i z e d i s a l s o measured by semantic p r o d u c t i v i t y . I t i s proposed here t h a t t h e r e are dimensions of semantic p r o d u c t i v i t y which are not o f t e n reached i n c u r r e n t language r e s e a r c h . 1 a. F o l l o w i n g the Syntax Paradigm Grammatical s p e c i f i c a t i o n has been faced with a s i m i l a r problem: - how to s p e c i f y a simple set of r u l e s which c h a r a c t e r i z e s the s y n t a c t i c p o s s i b i l i t i e s o f a language. The g r e a t success t h a t has been met i n grammatical a n a l y s i s has tempted r e s e a r c h e r s t o c a r r y over t h e i r t o o l s t o semantics. T h i s approach was f u r t h e r encouraged by the o b s e r v a t i o n t h a t the a b i l i t y of a l e x i c a l item to e n t e r i n t o a semantic r e l a t i o n was a property of a group of words r a t h e r than i n d i v i d u a l items. While these groups were more complex than t h e i r s y n t a c t i c c o u n t e r p a r t s , the same success was expected. The g u e s t i o n s to be answered by semantic a n a l y s i s could then be summarized as: (?1) How can l e x i c a l elements be assigned a t t r i b u t e s which t y p i f y t h e i r semantic c a p a b i l i t i e s ? (?2) How can r u l e s be devised which s p e c i f y the p o s s i b i l i t i e s f o r c o - e x i s t e n c e of l e x i c a l elements i n terms of these a t t r i b u t e s ? F o l l o w i n g the Syntax Paradigm Semantic P r o d u c t i v i t y 3 Attempts t o answer these q u e s t i o n s have been made by Katz and Fodor (1963) and Chomsky (1965), and c o n t r o v e r s y over the d e t a i l s of t h e i r s o l u t i o n s has raged ever s i n c e , B a s i c a l l y , t h e i r approach was to a s s i g n important c h a r a c t e r i s t i c s to each l e x i c a l element. T h i s i n f o r m a t i o n would then determine the p o i n t s a t which the word c o u l d be i n s e r t e d i n t o the grammatical s t r u c t u r e . There a r e , however, g l a r i n g problems with t h e i r approach, most of which b o i l down t o an i n a b i l i t y t o s p e c i f y the c r i t e r i a by which the semantic c h a r a c t e r i s t i c s are to be chosen ( B o l i n g e r , 1965; McCawley, 1968). Another t o t a l l y d i f f e r e n t concept of semantic s p e c i f i c a t i o n i s the use of semantic networks ( Q u i l l i a n , 1968; C o l l i n s and Q u i l l i a n , 1972). These systems r e q u i r e t h a t every p i e c e o f s t o r e d i n f o r m a t i o n i s a v a i l a b l e t o determine the meaning of an u t t e r a n c e . Knowledge was i n c l u d e d i n order to re p r e s e n t the g e n e r a l meanings of words, and not to r e p r e s e n t t h e i r use i n any s p e c i f i c s i t u a t i o n . Even the understanding of metaphor was c o n s i d e r e d to be a n a t u r a l outgrowth of t h i s approach t o the use o f knowledge { C o l l i n s and Q u i l l i a n , 1972:346). Thus these systems were s e m a n t i c a l l y p r o d u c t i v e , c o v e r i n g a wide range of p o s s i b l e word meanings without l i s t i n g these p o s s i b i l i t i e s e x p l i c i t l y . Dnf ortunately., however, t h i s method of s p e c i f i c a t i o n soon f e l l i n t o d i s f a v o u r due t o d i f f i c u l t i e s i n the d e f i n i t i o n and c o n t r o l of processes o p e r a t i n g upon the semantic network. With the e x c e p t i o n of the r e c e n t work of Bobrow and Winograd (1976), the r o l e of semantic networks has F o l l o w i n g the Syntax Paradigm Semantic P r o d u c t i v i t y 4 l a r g e l y been r e l e g a t e d to the r e p r e s e n t a t i o n of sentence meaning {see Schubert, 1976). 1b. Extending the Syntax Paradigm Two other q u e s t i o n s soon came i n t o f o c u s , l i k e l y i n response to the development of computational methods: (?3) How can the semantic r o l e of each component of a sentence be i d e n t i f i e d ? (?4) How can the r u l e s (?2) be organized to permit the semantic a n a l y s i s of a sentence? The f i r s t answer to these q u e s t i o n s was provided by Woods (1968), even though h i s work i s more o f t e n h a i l e d as having s o l v e d many of the s y n t a c t i c problems of computerized language a n a l y s i s . woods' p r o c e d u r a l semantics provided an o r g a n i z a t i o n o f semantic r u l e s by a l l o w i n g them to be attached t o any grammatical form i n the language s p e c i f i c a t i o n . The t h i r d q u e s t i o n was s i d e s t e p p e d i n t h a t the language a p p l i c a t i o n was question-answering, and the semantic r o l e of any component of the i n p u t was given i n terms of i t s e f f e c t on the s e a r c h of the data-base, Woods' system was, however, t y p i c a l of computer models i n t h a t i t almost completely ignored the f i r s t q u estion (?1) of how t o s p e c i f y semantic p r o p e r t i e s . A simple s e t of Extending the Syntax Paradigm Semantic P r o d u c t i v i t y 5 i n c l u s i o n p r o p e r t i e s was chosen, although he recognized t h a t t h e r e was a p o t e n t i a l need f o r a more complex s p e c i f i c a t i o n (1968:465), In the same year, F i l l m o r e presented h i s case f o r case (1968). T h i s 1 was c l e a r l y an attempt to answer g u e s t i o n s 3 and 4. The r o l e of each component i n a sentence was i d e n t i f i e d as one of a s e t of s p e c i f i e d c ases, and the case r u l e s were or g a n i z e d around the verbs. 1c. The Impact o f Case These new q u e s t i o n s which were being asked about semantics were of i n t e r e s t to those i n v o l v e d with computer models of language understanding because, r a t h e r than d e s c r i b i n g semantics, they emphasized procedures o p e r a t i n g on r e p r e s e n t a t i o n s . Most modern approaches to n a t u r a l language i n a r t i f i c i a l i n t e l l i g e n c e are based upon case systems (see the summaries of Wilks, 1974; Schubert, 1976; Bruce, 1973, 1975).. However, there are problems that have been i n h e r i t e d through adherence to t h i s approach. 1) What c o n s t i t u t e s a case? T h i s problem has been d i s c u s s e d i n d e t a i l by Bruce (1975). There seems t o be no c l e a r way to determine which a s p e c t s of the context surrounding the use of a word should be i n c l u d e d as case r o l e s . Attempts to The Impact of Case Semantic P r o d u c t i v i t y 6 s o l v e t h i s problem have u s u a l l y i n v o l v e d the e x t e n s i o n of the concept of case to i n c l u d e more and more about language use s i t u a t i o n s , f o l l o w e d by the p r o p o s a l of mechanisms to foreground the p a r t i c u l a r i l y r e l e v a n t i n f o r m a t i o n ( F i l l m o r e , 1976; G o l d s t e i n and Papert, 1976). 2) A case system r e q u i r e s the s p e c i f i c a t i o n of a s e t of p o s s i b l e semantic r o l e s t h a t a component (such as a p r e p o s i t i o n a l phrase) can play i n the meaning of a sentence. There has never been much agreement on the exact nature of t h i s s e t , but a new t r e n d i n p r i m i t i v i z a t i o n has been e s t a b l i s h e d . T h i s i s c a r r i e d over to the r e d u c t i o n of verbs i n t o a s m a l l s e t of p r i m i t i v e a c t i o n s by Schank (1972) and Norman and Rumelhart (1975). S i l k s (1975c) has shown t h a t the problems i n h e r e n t i n the use of p r i m i t i v e s are based on the f a l s e b e l i e f t h a t by using p r i m i t i v e s one can a v o i d the d i f f i c u l t i e s of using language to re p r e s e n t language (see McDermott, 1976). 3) The problem (?1) of the o r g a n i z a t i o n of semantic a t t r i b u t e s has been l a r g e l y ignored. In the next s u b - s e c t i o n the e x t e n t to which t h i s q u e s t i o n has been d e a l t with i n modern lanquage a n a l y s i s systems w i l l be demonstrated. The f o l l o w i n g s u b - s e c t i o n w i l l show some important f e a t u r e s of language use which have been ignored. The Impact of Case Semantic P r o d u c t i v i t y 7 1 d. The S p e c i f i c a t i o n o f Semantic A t t r i b u t e s Schank's (1972, 1975, 1976) system a s s i g n s a t t r i b u t e s t o each l e x i c a l element i n the same way t h a t K a t z and Fodor (1963) and Chomsky (1965) d i d (see Schank,1972:563). f o r example: "man" i s g i v e n the a t t r i b u t e "human" "book" i s g i v e n t h e a t t r i b u t e "physob" Each verb has a s p e c i f i c a t i o n of t h e s e m a n t i c c a t e g o r i e s t o w h i c h t h e s e n t e n c e components must belong ( i b i d :562). Thus a d i f f e r e n t sense o f a verb i s s p e c i f i e d f o r each d i f f e r e n t c a t e g o r y which can be a component. For example: "Dick p l a y e d t h e p i a n o " and " D i c k p l a y e d t h e song" would have t o be two d i f f e r e n t senses of "play"> u n l e s s both p i a n o and song c o u l d have some a t t r i b u t e such as " p l a y a b l e " . T h i s would r e p r e s e n t r e d u c t i o n t o a b s u r d i t y , and would r e s u l t i n v e r y l i t t l e s e m a n t i c p r o d u c t i v i t y , i n t h e sense t h a t each s e m a n t i c a l l y a c c e p t a b l e use of words would have i t s own p r i v a t e s e t o f s e m a n t i c r u l e s . Norman and Rumelhart (1972, 1973, 1975) d e v i s e d a system which s p e c i f i e s a case as a p r e d i c a t e , such a s : g i v e ( a g e n t , o b j e c t , r e c i p i e n t , t i m e ) The S p e c i f i c a t i o n o f Semantic A t t r i b u t e s Semantic P r o d u c t i v i t y 8 Each range (such as agent) posseses s e l e c t i o n a l r e s t r i c t i o n s on the concept t h a t can f i l l the p o s i t i o n (1975:41). Again, as with Schank's system, each sense of the verb i s represented by a s e p a r a t e p r e d i c a t e ( i b i d :41). Nouns and a d j e c t i v e s are r e p r e s e n t e d i n terms of p r o p e r t i e s such as: person (x) — > mortal (X) and biped (X) and a b l e (X, t a l k (X)) AND .... ( i b i d : 57) The verb p r e d i c a t e s c o n t a i n procedures which w i l l compute the goodness of f i t of any concept f o r a given range i n the p r e d i c a t e ( i b i d : 64) . The exact nature of these o p e r a t i o n s has been l e f t u n c l e a r (see a l s o Rumelhart and Norman, 1973). Another system has been e x p l i c i t l y based on case (Taylor and Rosenberg, 1975). T h i s system presented an i n d i c a t i o n t h a t s imple c a t e g o r i e s of l e x i c a l elements were not enough to determine the a c c e p t a b i l i t y of a l e x i c a l item as the case of a verb. Rather than s t r u c t u r i n g the a t t r i b u t e s of l e x i c a l elements, however, the concept of case was extended t o i n c l u d e a more complex mechanism f o r the e v a l u a t i o n of p o t e n t i a l cases. Each verb "knew" the necessary, most l i k e l y and p o s s i b l e semantic a t t r i b u t e s t h a t determine the a c c e p t a b i l i t y of a l e x i c a l item. To summarize, case-based systems have used only b a s i c s emantic-marker-like p r o p e r t i e s o f words to determine t h e i r acceptance as c a s e - f i l l e r s . Attempts to extend c a s e - f i l l i n g 0 The S p e c i f i c a t i o n of Semantic A t t r i b u t e s Semantic P r o d u c t i v i t y 9 a b i l i t i e s have been centered around the development of more complex p r o c e d u r a l i n f o r m a t i o n a s s c o c i a t e d with each case-frame (see a l s o G o l d s t e i n and P a p s r t , 1976). This approach has not i n c r e a s e d the semantic p r o d u c t i v i t y of the systems i n t h a t v a r i a t i o n s i n shades of meaning become e f f e c t i v e l y d i f f e r e n t l e x i c a l elements. Two other i n t e r e s t i n g d i r e c t i o n s are worth n o t i n g here. Winograd (1971:302) uses K a t z - F o d o r - l i k e semantic markers to f i l t e r out meaningless i n t e r p r e t a t i o n s , but a l s o i n c l u d e s an a b i l i t y to examine the " r e a l world" i n order t o r e s o l v e p o t e n t i a l l y ambiguous i n p u t s . T h i s approach i s e f f e c t i v e w i t h i n a c o n s t r a i n e d a p p l i c a t i o n i n which i n f o r m a t i o n conveyed i n the sentences concerns observable phenomena. Anderson and Bower (1972:192-196) propose the i d e a of a " d e f i n i n g use" of a concept. T h i s a r c h e t y p a l use of a word p r o v i d e s s e l e c t i o n a l r e s t r i c t i o n s i n terms of the semantic c a t e g o r i e s of the words which appear with the d e f i n e d word. While t h i s approach a l l o w s the system to l e a r n new word-uses, i t i s l a c k i n g i n semantic p r o d u c t i v i t y i n t h a t each s l i g h t shade of meaning would r e q u i r e i t s own " d e f i n i n g use". 1e. M i l k s ' Approach A case-based system that Wilks (1972, 1973, 1975a, 1975b) has designed i s d i s t i n g u i s h e d f o r t h r e e reasons. Wilks' Approach Semantic P r o d u c t i v i t y 10 F i r s t , the system provides a h i e r a r c h y of mechanisms f o r the s p e c i f i c a t i o n and determination of the semantic r o l e s of the components of a sentence. ; Second, the system s p e c i f i e s a category ( p r i m i t i v e ) f o r each l e x i c a l element so t h a t the e n t i r e s e t of s e m a n t i c a l l y v a l i d sentences can be given i n terms of these p r i m i t i v e s (as b a r e - t e m p l a t e s ) . For example: MAN CAUSE SIGN i s a bare template which may be used to rec o g n i z e the sentence "the crook sang a song" The important p o i n t here i s not the d e t a i l s o r the e f f e c t i v e n e s s o f Wilks' s p e c i f i c a t i o n s , but the r e a l i z a t i o n t h a t semantic c a t e g o r i e s and the a t t r i b u t e s f o r l e x i c a l elements must be designed together with the case-frame s p e c i f i c a t i o n s . T h i r d , Wilks' system provides an a d d i t i o n a l degree of f l e x i b i l i t y i n the a p p l i c a t i o n of the cases. L e x i c a l elements are given p r i m i t i v e c a t e g o r i e s , and the p r i m i t i v e s themselves can belong t o c a t e g o r i e s at a higher l e v e l . For example "crook" belongs to the category MAN, which i n turn belongs to *ANIMATE, hence "crook" can f i l l a case f o r a verb which r e q u i r e s an •ANIMATE s u b j e c t . T h i s aspect of an o r g a n i z a t i o n of semantic a t t r i b u t e s i s extended i n the model proposed i n a l a t e r s e c t i o n . Another f l e x i b i l i t y i n the use of semantic a t t r i b u t e s i s demonstrated i n only one plac e i n Wilks' work (1974:20). In w i l k s ' Approach Semantic P r o d u c t i v i t y t h i s example i t i s shown t h a t although " d r i n k " r e q u i r e s a " l i q u i d " o b j e c t , that "he drank the p i t c h e r " would be a c c e p t a b l e , because the s p e c i f i c a t i o n of the word " p i t c h e r " i n c l u d e d mention of the concept " l i q u i d " . T h i s aspect of semantic p r o d u c t i v i t y w i l l be expanded and f o r m a l i z e d i n a l a t e r s e c t i o n . . 1f. what Has Been Ignored With the e x c e p t i o n of a few h i n t s given i n Wilks' work a l l o f the systems d i s c u s s e d so f a r have ignored two important i d e a s . 1) There are i n t r i c a t e semantic r e l a t i o n s among l e x i c a l elements. 2) These r e l a t i o n s decide the a c c e p t a b i l i t y of the words as case f i l l e r s . C o n sider again the verb "to p l a y " i n the sense that humans mus i c a l instruments. One c o u l d d e s c r i b e a case-system f o r verb as f o l l o w s : play ( s u b j e c t , o b j e c t ) i n which s u b j e c t : human o b j e c t : m u s i c a l instrument but t h i s would r u l e out sentences such as What Has Been Ignored Semantic P r o d u c t i v i t y 12 i) "Dick played the Moonlight Sonata" which i s c l e a r l y an example o f the same sense of the verb. I t i s the e x i s t e n c e of a r e l a t i o n between "musical-instrument" and "Moonlight Sonata" that allows i t to take the place of "musical i n s t r u m e n t " , The sentence: i i ) "Dick played the piano" i s a c c e p t a b l e because "piano" i s an i n s t a n c e of "musical instrument". Most systems r e c o g n i z e the use of t h i s semantic r e l a t i o n , but sentence i ) i s a c c e p t a b l e because: humans play musical instruments AND musical instruments emit music AND Moonlight Sonata i s music. Not j u s t any r e l a t i o n between concepts w i l l allow the s u b s t i t u t i o n of l e x i c a l elements. For example: "Dick played the f a c t o r y " i s not a c c e p t a b l e j u s t because musical instruments are made i n f a c t o r i e s . To r e a l i z e the complexity of the r u l e s i n v o l v e d i n a l l o w i n g such a c a s e - r o l e t o be f i l l e d , c o n s i d e r the p o s s i b l e o b j e c t s o f the verb " t o p l a y " ( i n the same s e n s e ) . Included a r e : musical instruments, piano, grand piano, Beethoven, the c l a s s i c s , the b l u e s . Moonlight Sonata, the song, the s t a n z a , the chord, the note, the s c a l e , the bass, the t r e b l e , the keyboard. fthat Has Been Ignored Semantic P r o d u c t i v i t y 13 the s t r i n g s , f sharp, a s e t , a c o n c e r t , The R i t z H o t e l , a wedding, p e r c u s s i o n , the woodwinds, (? the r e c o r d p l a y e r ) , ( ? the tape r e c o r d e r ) . T h i s s e c t i o n i s t i t l e d "Semantic P r o d u c t i v i t y " . I t i s the c o n t e n t i o n of t h i s paper t h a t : 1) A more powerful semantic p r o d u c t i v i t y can be reached by c o n s i d e r i n g the o r g a n i z a t i o n of semantic r e l a t i o n s among the l e x i c a l elements, r a t h e r than r e l y i n g s o l e l y on the " i s " r e l a t i o n . 2) As Wilks has p o i n t e d out, the nature o f these r e l a t i o n s should be r e f l e c t e d i n the choice of case systems. The case f i n d i n g p rocessor should have access t o , and understand the o r g a n i z a t i o n of semantic r e l a t i o n s among the l e x i c a l e n t r i e s . The i n t r o d u c t i o n of new i n f o r m a t i o n should not f o r c e a r e c o n s i d e r a t i o n of case-frames f o r verbs which could use the i n f o r m a t i o n . T h i s can be accomplished i n a s e m a n t i c a l l y p r o d u c t i v e system which l o c a l i z e s the use of any word and permits f l e x i b i l i t y i n determining the a c c e p t a b i l i t y of i t s c a s e - f i l l e r s . What Has Been. Ignored O b j e c t i v e s and Overview 14 2. O b j e c t i v e s and Overview In g e n e r a l , the o b j e c t i v e of t h i s work i s to devise a s e m a n t i c a l l y p r o d u c t i v e method f o r the r e p r e s e n t a t i o n of semantic knowledge to be used i n language a n a l y s i s . The approach i s to demonstrate the combined use of case-frames and semantic networks, not i n the r e p r e s e n t a t i o n of meaning but i n the d i s c o v e r y of meaning. I t i s expected t h a t : 1) The methods f o r determining the c a s e - f i l l i n g a b i l i t i e s of nouns can be extended beyond semantic-markers to i n c l u d e i n f o r m a t i o n c o n t a i n e d i n a semantic network s t r u c t u r e . 2) The r e l a t i o n s among verbs which are expressed i n the semantic network can be used to d i r e c t and c o n t r o l the process of s e a r c h i n g f o r case frames to be s a t i s f i e d . The more s p e c i f i c goals can best be expressed f o l l o w i n g a b r i e f o u t l i n e of the proposed system. 2a. B r i e f O u t l i n e The main f e a t u r e s of the model a r e : 1) A knowledge base which c o n t a i n s p i e c e s of i n f o r m a t i o n ranging from s p e c i f i c to very g e n e r a l knowledge. 2) A processor which attempts to l o c a t e those p i e c e s of B r i e f O u t l i n e O b j e c t i v e s and Overview 15 i n f o r m a t i o n which are r e l e v a n t to the understanding of a given i n p u t . Very l i t t l e s t r u c t u r a l r e l a t i o n i s proposed among these i d e n t i f i e d components, but each p a r t of the i n p u t sentence i s r e l a t e d to s p e c i f i c knowledge p i e c e s . Any f u r t h e r o r g a n i z a t i o n of the knowledge p i e c e s i s assumed t o be dependent upon the s p e c i f i c language-use being modeled. The knowledge base c o n s i s t s of i n d i v i d u a l p i e c e s of i n f o r m a t i o n . These p i e c e s are termed prototypes because they r e p r e s e n t , or were used t o form r e p r e s e n t a t i o n s o f , language use ex p e r i e n c e s t o which the comprehension process appeals to be a b l e to understand a new u t t e r a n c e . These prototypes are rep r e s e n t e d u s u a l l y as s u b j e c t - v e r b - o b j e c t t r i p l e s , with other p o s s i b l e attachments. T h i s i s a s i m p l i f i c a t i o n which r e q u i r e s f u r t h e r explanation.,. I t i s the i n t e n t of t h i s paper to demonstrate an o r g a n i z a t i o n o f semantic knowledge, and the way t h i s knowledge can be used i n comprehension. As a r e s u l t , a simple knowledge r e p r e s e n t a t i o n i s being proposed which w i l l permit t h a t demonstration without too much c o m p l i c a t i o n . The s t r u c t u r e of the r e p r e s e n t a t i o n of a prototype w i l l be simple s u b j e c t , verb, o b j e c t p r o p o s i t i o n s . A sentence i n p u t t o the system w i l l a l s o be i n a s i m i l a r from. T h i s s i m p l i f i c a t i o n s h ould not be i n t e r p r e t e d as a b e l i e f t h a t grammatical knowledge i s t r i v i a l and can be e x t r a c t e d by p r e - p r o c e s s i n g . The B r i e f O u t l i n e O b j e c t i v e s and Overview 16 r e s t r i c t i o n a l l o w s a f o c u s s i n g on the problems of semantic a n a l y s i s s e p a r a t e l y from complex grammatical c o n s i d e r a t i o n s . The processes which operate over the knowledge base are concerned with the comprehension of an i n p u t sentence. The a c t i v e attempt t o i d e n t i f y p i e c e s of knowledge (prototypes) w i t h i n the knowledge-base which have something t o do with the i n p u t sentence i s c a l l e d resolution. A c t u a l l y , r e s o l u t i o n t r i e s t o f i n d the prototypes that have "the most" t o do with the i n p u t . The system w i l l always attempt to make the best sense p o s s i b l e from the i n p u t , and i s not expected t o be d e a l i n g with i l l - f o r m e d sentences. I f a sentence i s given to the system which i s "meaningless" i n the sense t h a t i t does not r e l a t e t o any knowledge t h a t the system has, then c l e a r l y i t w i l l not i d e n t i f y any prototypes i n the process of r e s o l u t i o n . Another p o i n t about the e x p e c t a t i o n s of the system should be made here. The prototypes which are s t o r e d i n the knowledge base are taken t o represent a l l of the knowledge t h a t the model has about language use. For example, i f prototypes e x i s t f o r two p i e c e s o f knowledge. "Elephants l i v e i n A f r i c a and have t u s k s . " "Rhinoceroses l i v e i n A f r i c a and have t u s k s . " The system (or a human) cannot be expected, on the b a s i s of t h i s knowledge alone to know that d i f f e r e n t tusks but the same A f r i c a are intended. T h i s , and other problems of the completeness and B r i e f O u t l i n e O b j e c t i v e s and Overview 17 c o n s i s t e n c y of knowledge are not being c o n s i d e r e d (see HcDermott, 1976). The system i s not expected to make d e c i s i o n s on the b a s i s of knowledge that i t does not have. 2b. Some S p e c i f i c O b j e c t i v e s Much more w i l l be s a i d about the nature of the components o f the system and t h e i r o r g a n i z a t i o n i n a l a t e r s e c t i o n . Some of the s p e c i f i c requirements which have d i r e c t e d the development o f the system a r e : 1) To r e f l e c t the h i e r a r c h i c a l nature of semantic knowledge and i t s a p p l i c a t i o n . 2) To demonstrate .a smooth continuum between s p e c i f i c and g e n e r a l i z e d semantic knowledge. 3) To show the r o l e of cases i n the o r g a n i z a t i o n and a p p l i c a t i o n of knowledge. 4) To devise a system f o r the s p e c i f i c a t i o n of the p r i m i t i v e s w i t h i n which knowledge i s to be stored which i s j u s t i f i e d by, c o n s i s t e n t with, and developed in r e l a t i o n t o both the h i e r a r c h i c a l s t r u c t u r e and the case systems i n use. 5) To re p r e s e n t semantic knowledge i n such a way t h a t language l e a r n i n g might be cons i d e r e d . Some S p e c i f i c O b j e c t i v e s Some Ex p l a n a t i o n s 18 3. ...Some Ex p l a n a t i o n s There are some " o u t e r - l e v e l " assumptions made which r e q u i r e e x p l a n a t i o n . These e x p l a n a t i o n s are necessary i n order t o b r i n g to the s u r f a c e the u n d e r l y i n g b e l i e f s about the d i r e c t i o n t h a t n a t u r a l language p r o c e s s i n g should be t a k i n g . I t i s because o f these b e l i e f s that the proposed model has been developed, and they are important i n determining the place t h a t t h i s type of r e s e a r c h might take i n the f i e l d of n a t u r a l language study. 3a. A Two-Process View of Comprehension One fundamental b e l i e f that i s r e f l e c t e d i n t h i s work i s t h a t i t i s worthwhile s e a r c h i n g f o r a l e v e l o f language a n a l y s i s which i s independent of the p a r t i c u l a r v e r b a l behavior being modeled. Language i s o f t e n s t u d i e d by examining s p e c i f i c v e r b a l a b i l i t i e s . Some of the forms of these a b i l i t i e s are: - paraphrase (Schank, 1975:238) - guestion answering (Charniak, 1975:46-48) - i n t e r p r e t a t i o n of commands (winograd,1972) - r e s o l u t i o n of r e f e r e n c e ( B i l k s , 1975a; Nash-Webber, 1976) - t r a n s l a t i o n ( C a r r o l l , 1972:23) (Wilks, 1973, 1975a) - p r e d i c t i o n (Davidson, 1976) A Two-Process View of Comprehension Some Exp l a n a t i o n s 19 N a t u r a l language systems share l i t t l e i n t h e i r approaches t o p r o c e s s i n g . Each system emphasizes those aspects of language which are of p a r t i c u l a r importance to the s p e c i f i c behavior being c o n s i d e r e d . I t seems unreasonable t h a t language p r o c e s s i n g i s to become a s e r i e s of s p e c i a l i z e d programs to d e a l with i n d i v i d u a l language-use s i t u a t i o n s , or a t l e a s t not a c c e p t a b l e u n t i l a l l avenues are i n v e s t i g a t e d i n search of some common bottom-up component (see Wilks, 1976 f o r a d i s c u s s i o n of t h i s p o i n t ) . I t appears t h a t a l l human knowledge and a b i l i t i e s w i l l e v e n t u a l l y be r e q u i r e d i n order to model n a t u r a l language understanding i n i t s e n t i r e t y . However, i t may be the case t h a t t h e r e are l e v e l s of comprehension which r e q u i r e only manageable p o r t i o n s of human knowledge., L i t t l e c o r r e l a t i o n e x i s t s among human c a p a b i l i t i e s i n the areas of v e r b a l behavior l i s t e d above, s u g g e s t i n g t h a t d i s t i n c t processes are i n v o l v e d ( C a r r o l l , 1972:3), The only s i g n i f i c a n t r e l a t i o n i n v o l v i n g these a b i l i t i e s i s t h a t they are h i g h l y c o r r e l a t e d to "high l e v e l " , and even non-verbal, i n t e l l e c t u a l a b i l i t i e s " ( C a r r o l l , 1972). I t i s o f t e n assumed t h a t the process of i n f e r e n c e must always be c o n s i d e r e d when d e a l i n g with language comprehension (Schank, 1976:163), and requirements f o r t h i s a b i l i t y w i l l vary depending on the p a r t i c u l a r v e r b a l behavior being s t u d i e d . To t r a n s l a t e and to answer q u e s t i o n s may i n v o l v e d i f f e r e n t demands on t h i s p r o c e s s . Consider the sentence & Two-Process View of Comprehension Some E x p l a n a t i o n s 20 "John r e c e i v e d a q u a r t e r from B i l l and gave i t t o Mary." C e r t a i n l y i n f e r e n c e i s r e q u i r e d to r e a l i z e t h a t John had no net gai n or l o s s i n the t r a n s a c t i o n , but i s t h a t an i n t e g r a l part of understanding the sentence, or i s i t a secondary process i n comprehension, t r i g g e r e d by a lower l e v e l understanding of the sentence? C l e a r l y the sentence could be e i t h e r paraphrased or t r a n s l a t e d without making the i n f e r e n c e . The model proposed i n t h i s paper i m p l i e s a two-process view of comprehension ( C a r r o l l , 1972) i n v o l v i n g : 1) The i d e n t i f i c a t i o n of p o s s i b l y r e l e v a n t knowledge. 2) The a p p l i c a t i o n o f t h i s i n f o r m a t i o n t o a wider context d i r e c t e d by the g o a l s and requirements o f experiments, or o b j e c t i v e s of models and implementations. T h i s l e v e l of comprehension i s not demonstrable, and should not be thought of as corresponding t o any r e a l d i v i s i o n i n human p r o c e s s i n g . However, t o p o s t u l a t e such a d i s t i n c t i o n may pr o v i d e p r o d u c t i v e i n s i g h t i n t o the problems of n a t u r a l language, p a r t i c u l a r l y i f a s u c c e s s f u l implementation can be based on t h i s d i s t i n c t i o n , and i f new outlooks on other work i n the f i e l d can be generated from w i t h i n t h i s i n t e r p r e t a t i o n . C e r t a i n l y the s p e c i f i c language-use c o n t e x t must e x e r t a top-down i n f l u e n c e on p r o c e s s i n g at a l l l e v e l s . The d i v i s i o n i n p r o c e s s i n g proposed here i s only a conceptual d i v i s i o n , to A Two-Process View of Comprehension Some Ex p l a n a t i o n s 21 permit a l e v e l at which some aspects of language comprehension can be analyzed as a bottom-up system which accesses l i m i t e d amounts of knowledge and which i s not i n e x t r i c a b l y bound t o a s p e c i f i c context or behavior. 3b. Comprehension Without Representation The proposed system has no d i s t i n c t r e p r e s e n t a t i o n of the meaning of an in p u t sentence. T h i s should not be i n t e r p r e t e d as a s i m p l i f i c a t i o n . The r e p r e s e n t a t i o n o f meaning has b e n e f i t s f o r some asp e c t s o f language understanding, but i t i s the c o n t e n t i o n expressed i n t h i s s e c t i o n t h a t over-emphasis on forms o f r e p r e s e n t a t i o n can l e a d t o an un n a t u r a l approach to the study o f n a t u r a l language. The most obvious e f f e c t of the s h i f t towards s t u d y i n g meaning r e p r e s e n t a t i o n has been the corresponding l o s s of i n t e r e s t i n studying the knowledge necessary t o accomplish language comprehension. Powerful mechanisms have been developed f o r the r e p r e s e n t a t i o n of t h i s knowledge ( F i l l m o r e , 1968; Q u i l l i a n , 1968), but r e c e n t l y these r e s u l t s have been i n t e r p r e t e d as t o o l s to demonstrate the i n t e r a c t i o n of components of a meaning r e p r e s e n t a t i o n (Bruce, 1975; Schubert, 1976). As a r e s u l t , semantic knowledge i s o f t e n l i t t l e more than p a r t i a l l y completed segments of the meaning s t r u c t u r e to be c r e a t e d (see f o r example Woods' (1968) templates, Wilks' (1975a) bare templates, or Schank's (1972) con c e p t u a l d i c t i o n a r y ) . Comprehension Without Repr e s e n t a t i o n Some Ex p l a n a t i o n s 22 There are t h r e e other d i r e c t r e s u l t s of t h i s approach: 1) I t has become d e s i r a b l e to r e p r e s e n t and use knowledge of s p e c i f i c events and s i t u a t i o n s (Schank, 1975b; Winograd, 1972). Since the semantic knowledge i s r e p r e s e n t e d as components of the meaning r e p r e s e n t a t i o n , t h i s s p e c i f i c knowledge must be s t o r e d s e p a r a t e l y , and be used by separate processes. I t i s not p o s s i b l e t o draw the l i n e between r e a l - w o r l d (e p i s o d i c ) knowledge and semantic (language-use) knowledge. The f a c t t h at they are kept separate has caused d i f f e r e n c e s i n the approaches to the o r g a n i z a t i o n of knowledge f o r language a n a l y s i s . (see f o r example the d i f f e r e n c e s between (Schank, 1976) and ( G o l d s t e i n and Papert, 1976)). F i l l m o r e (1974) has po i n t e d out how untenable t h i s d i s t i n c t i o n i s f o r any r e a l i s t i c approach to language comprehension. 2) From w i t h i n t h i s paradigm, language-use knowledge cannot r e c e i v e i n f o r m a t i o n fed-back from the meaning of sentences, and hence l a n g u a g e - l e a r n i n g i s s p e c i f i c a l l y excluded, ft r e a l i s t i c view o f language comprehension i s t h a t understanding i n v o l v e s the a s s i m i l a t i o n of i n f o r m a t i o n i n t o our e x i s t i n g knowledge. T h i s should not be i n t e r p r e t e d as an o b j e c t i o n to the p s y c h o l o g i c a l v a l i d i t y o f e x i s t i n g language understanding models, but r a t h e r as an a s s a u l t on the concept of "understanding", whether i n machine or human. The o b j e c t i v e should be t o c h a r a c t e r i z e the methods by which knowledge can be absorbed v i a language, r a t h e r than how the Comprehension Without R e p r e s e n t a t i o n Some Ex p l a n a t i o n s 23 "meaning" of a s p e c i f i c sentence could be presented. 3) The l a c k of i n t e r e s t i n the s t r u c t u r e of semantic knowledge has l e d to a concern f o r more s p e c i f i c knowledge about events and s i t u a t i o n s which could be used to guide the language a n a l y s i s process, hence there has been a tendency towards d e a l i n g with s m a l l e r domains, and the use of formalisms which are not e a s i l y extended t o oth e r , p a r t i c u l a r l y l a r g e r , domains. The form of a meaning r e p r e s e n t a t i o n i s h e a v i l y i n f l u e n c e d by the p a r t i c u l a r aspect of language which i s being s t u d i e d . Thus t h i s s h i f t i n emphasis has f u r t h e r e d a d i s p a r i t y among, and i s o l a t i o n o f , r e s e a r c h e f f o r t s i n n a t u r a l language. The use of meaning r e p r e s e n t a t i o n as the b a s i s f o r e x t r a c t i n g i n f o r m a t i o n and drawing i n f e r e n c e s has l e d to an assumption about a f i x e d l e v e l of knowledge a v a i l a b i l i t y i n the system. Together with each a c t i o n expressed i n an in p u t sentence there i s o f t e n an a s s o c i a t e d f i x e d l e v e l of c a u s a l and phsnomenological i n f e r e n c e s (Schank, 1976). For any p i e c e of knowledge, i t i s p o s s i b l e t o devise a sentence which cannot be understood without t h a t knowledge. I t seems unreasonable t h a t some knowledge i s always considered as part of an a c t i o n , and t h a t other knowledge i s not. Consider the f o l l o w i n g sentences: (i) John d r i n k s g i n again now t h a t he's had h i s u l c e r s Comprehension Without Representation Some Ex p l a n a t i o n s 24 t r e a t e d . ( i i ) John d r i n k s lemon gin with orange soda. In the f i r s t sentence, the knowledge t h a t d r i n k i n g i n v o l v e s i n g e s t i o n to the stomach i s very important, and yet t h a t i s i r r e l e v a n t t o the second sentence. That humans can p r a g m a t i c a l l y extend or cut o f f the depth o f a n a l y s i s , seems to be an important component o f language understanding, which has been l a r g e l y ignored through a d e s i r e to o b t a i n a f i x e d l e v e l of r e p r e s e n t a t i o n f o r meaning. Once the b a s i c i n f o r m a t i o n has been e x t r a c t e d from an u t t e r a n c e , and some modeled behavior i s d e s i r e d , then the f o r m a t i o n of a meaning r e p r e s e n t a t i o n may be necessary i n order t h a t proper planning and i n f e r e n c e s can be made. Comprehension Without Representation A Knowledge I d e n t i f i c a t i o n System 25 4..A Knowledge I d e n t i f i c a t i o n System The b r i e f o u t l i n e of the system, provided i n an e a r l i e r s e c t i o n , suggested a data-base of pieces of i n f o r m a t i o n from which r e l e v a n t knowledge co u l d be i d e n t i f i e d f o r any input sentence., The theory necessary t o organize t h i s i n f o r m a t i o n and c o n t r o l the processes of i d e n t i f i c a t i o n w i l l be e x p l a i n e d i n t h i s s e c t i o n . The approach which has been taken i s to g r a d u a l l y develop the theory. F i r s t , a very b a s i c v e r s i o n w i l l be presented, c o n s i s t i n g of a simple process of r e s o l u t i o n and a t r i v i a l o r g a n i z a t i o n of prototypes. Each aspect which i s i n t r o d u c e d does, however, remain w i t h i n the more complicated v e r s i o n s . A d d i t i o n a l f e a t u r e s w i l l be e x p l a i n e d as they are added u n t i l , u n f o r t u n a t e l y , a s p e c t s of the theory become h i g h l y s p e c u l a t i v e . 4a. The B a s i c Version Consider a system which c o n s i s t s o f the f o l l o w i n g p r o t o t y p e s : Tom d r i n k s milk Sam d r i n k s t e a Tom i s human Sam i s human John i s human Milk i s l i q u i d Tea i s l i q u i d C o ffee i s l i q u i d The B a s i c V e r s i o n A Knowledge I d e n t i f i c a t i o n System 26 These p r o t o t y p e s r e p r e s e n t knowledge t h a t the system has i n terms o f language e x p e r i e n c e s which i t has p r e v i o u s l y e n c o u n t e r e d . Each p r o t o t y p e has t h r e e Components, whose g r a m m a t i c a l r o l e s a r e o b v i o u s and are not i n d i c a t e d . An o r g a n i z a t i o n i s assumed by which any l e x i c a l element has a c c e s s t o a l l of the p r o t o t y p e s i n which i t i s i n v o l v e d . The o b j e c t o f the p r o c e s s of r e s o l u t i o n i s t o l o c a t e a p r o t o t y p e which matches the i n p u t g r a m m a t i c a l l y ( f o r now they are a l l t h e same) and whose components may be traced from t h e components of t h e i n p u t s e n t e n c e . A component o f an i n p u t s e n t e n c e t r a c e s t o a component o f t h e t a r g e t p r o t o t y p e i f the components a r e t h e 'same, or i f a n o t h e r p r o t o t y p e can be l o c a t e d which c o n t a i n s the i n p u t component, and which a l s o c o n t a i n s a n o t h e r component which t r a c e s t o the t a r g e t component. For example: John d r i n k s c o f f e e ( i n p u t ) * • * • • * • • • John i s human . C o f f e e i s l i q u i d • • • Tom i s human . M i l k i s l i q u i d Tom d r i n k s m i l k I n t h i s example, "John d r i n k s c o f f e e " i s the i n p u t . The p r o t o t y p e "Tom d r i n k s m i l k " i s a s u c c e s s f u l t a r g e t f o r r e s o l u t i o n because a l l o f i t s components can be t r a c e d from the i n p u t components. The verbs t r a c e d i r e c t l y . The s u b j e c t , " J o h n " t r a c e s t o "Tom" t h r o u g h t h e e x i s t e n c e of a n o t h e r The B a s i c V e r s i o n A Knowledge I d e n t i f i c a t i o n System 27 prototype "John i s human", which has a component "human" that traces to "Tom". This second step i n the tracing i s again accomplished by the existence of another prototype "Tom i s human". The objects are traced i n a si m i l a r way. The conditions for tracing can be expressed more formally as: Let T[p] represent that T i s a component of prototype p. Also, l e t S—>T [ p J with X represent "S traces to T (in prototype p), using prototypes X", then for the complete set of prototypes Z, S—>T[p] with Z«=Z-p i f f S=T or (3g€Z«) S6g & (3R6g) R*S S R—>T[p] with Z'-g where "=" indicates i d e n t i t y . There are several observation about the system which can be made at th i s point: 1) A measure of resolution can be derived from the length of the traces of the prototypes's components, and t h i s measure could be used to decide among competing resolutions. 2) Often more than one choice of prototype can be made, even with the same measure of resolution. In the above example, "Sam drinks tea" could have been selected. The Basic Version A Knowledge I d e n t i f i c a t i o n System 28 3) A system c o u l d not r e a s o n a b l y examine e v e r y p i e c e o f i n f o r m a t i o n i n o r d e r t o s e l e c t t h o s e t h a t a r e r e l e v a n t t o the i n p u t . W i t h i n t h e framework of t h e c u r r e n t v e r s i o n , two p o s s i b i l i t i e s e x i s t . F i r s t , more r i g i d c o n d i t i o n s c o u l d be imposed on t h e t r a c i n g o f v e r b s , f o r i n s t a n c e , t h a t they be i d e n t i c a l (as i n t h e c a s e o f t h e example). T h i s would g r e a t l y r e s t r i c t t h e s e t of p r o t o t y p e s t o be c o n s i d e r e d . S e c o n d l y , t h e s e a r c h f o r a r e s o l u t i o n might be g u i d e d by t h r e s h o l d s on t h e a c c e p t a b i l i t y o f the measure of r e s o l u t i o n . Such t h r e s h o l d s c o u l d i n s u r e t h a t t r a c i n g would o n l y be attempted t o a r e a s o n a b l e l e n g t h , and t h a t once a p r o t o t y p e i s r e s o l v e d w i t h a low enough measure, t h e e n t i r e p r o c e s s c o u l d s t o p . These t e c h n i q u e s have been i n c o r p o r a t e d i n t o l a t e r v e r s i o n s o f the t h e o r y and are a p a r t of t h e implemented model t o be d i s c u s s e d i n s e c t i o n s i x . 4b. H i e r a r c h i e s And G e n e r a l i z a t i o n s I n t h e b a s i c model, each p r o t o t y p e r e p r e s e n t s v e r y s p e c i f i c i n f o r m a t i o n . The n e x t s t e p i n t h e development o f the system i s t o p e r m i t a more g e n e r a l form o f knowledge t o r e p r e s e n t s e v e r a l s p e c i f i c i n s t a n c e s . For example "humans d r i n k l i q u i d s " would s u f f i c e t o c o v e r the t h r e e p r o t o t y p e s about " d r i n k i n q " i n the b a s i c model. T h i s form of p r o t o t y p e , termed a generalization, cannot e x i s t by i t s e l f , t h e r e must be a method of s p e c i f y i n g the H i e r a r c h i e s And G e n e r a l i z a t i o n s A Knowledge I d e n t i f i c a t i o n System 29 c r i t e r i a f o r i t s use i n r e s o l v i n g an i n p u t . Thus, h i e r a r c h i c a l r e l a t i o n s a r e proposed which w i l l i n d i c a t e t h o s e more s p e c i f i c e l e m e n t s which may be used i n t h e r e s o l u t i o n i n v o l v i n g a g e n e r a l i z e d p i e c e o f knowledge. C o n s i d e r the f o l l o w i n g example: Human d r i n k l i q u i d Tom <human> Sam <human> John <human> C o f f e e < l i q u i d > M i l k < l i g u i d > Tea < l i g u i d > The r e p r e s e n t a t i o n A<B> has been used t o i n d i c a t e t h a t B has been s e t h i q h e r than A i n a h i e r a r c h i c a l r e l a t i o n s h i p . I n the example, o n l y one p r o t o t y p e r e m a i n s , a l l o t h e r s have e i t h e r been subsumed by i t o r c o n v e r t e d i n t o h i e r a r c h i c a l r e l a t i o n s h i p s . I t i s i n t e r e s t i n q t o note t h a t t h e depth of t h e s e r e l a t i o n s seldom exceeds f o u r l e v e l s ( C o l l i n s and Q u i l l i a n , 1972:321). The r e q u i r e m e n t s f o r the t r a c i n g of p r o t o t y p e components has now been extended t o i n c l u d e t h e p o s s i b i l i t y o f u s i n g such h i e r a r c h i c a l r e l a t i o n s h i p s . A q a i n , more f o r m a l l y : S — > T [ p ] w i t h Z»=Z-p i f f S=T or S<T> o r (3qez«) S€q s QR€q) R#S & R—>T[p] withZ»-q example o f such a r e s o l u t i o n would be: H i e r a r c h i e s And G e n e r a l i z a t i o n s A Knowledge I d e n t i f i c a t i o n System 30 John d r i n k s c o f f e e (input) John <human> , Co f f e e <liquid> V * Human d r i n k s l i q u i d (prototype) There i s nothing new i n the use of such h i e r a r c h i e s to e s t a b l i s h the meaning of i n p u t sentences. As was pointed out i n an e a r l i e r s e c t i o n , most n a t u r a l language systems r e l y s o l e l y on such mechanisms t o determine the semantic r o l e s of l e x i c a l elements. There a r e , however, advantages i n e s t a b l i s h i n g t h i s f e a t u r e i n the c u r r e n t framework. In order to c l a r i f y the f i r s t advantage, i t i s necessary to p o i n t out the s i m i l a r i t i e s between the system as i t stands now, and a case-system. The g e n e r a l i z e d p r o t o t y p e s a re c a s e - l i k e c o n s t r u c t i o n s f o r verb use, and the h i e r a r c h i c a l o r g a n i z a t i o n of nouns r e p r e s e n t s the a t t r i b u t e s which allow them t o f i l l the case r o l e s . Of course, at t h i s p o i n t the system only d e a l s with s u b j e c t and o b j e c t cases, but i n a l a t e r s e c t i o n i t w i l l be shown t h a t a n a t u r a l e x t e n s i o n can be made to i n c l u d e those r o l e s u s u a l l y encountered i n a case-system. The advantage f o r t h i s system i s t h a t , r e g a r d l e s s of i t s p a r t of speech, any l e x i c a l item can access the prototypes i n H i e r a r c h i e s And G e n e r a l i z a t i o n s A Knowledge I d e n t i f i c a t i o n System 31 which i t i s i n v o l v e d . This p r o v i d e s an i n c r e a s e i n power. Knowledge other than ISA r e l a t i o n s h i p s can be used t o determine case r o l e a c c e p t a b i l i t y , and at the same time, each pi e c e of i n f o r m a t i o n e x i s t s independently of the way i t i s t o be used. Consider the example: John drank the b o t t l e (input) • * • • * • * • • John <human> . B o t t l e s c o n t a i n l i q u i d • • • • • • • • • # t * Humans d r i n k l i q u i d s (prototype) The component "John" i s t r a c e d t o "human" through i t s h i e r a r c h i c a l r e l a t i o n s , and " b o t t l e " i s t r a c e d through i t s involvement i n a prototype with " l i q u i d " . Note t h a t the prototype " b o t t l e s c o n t a i n l i q u i d " c o u l d be used as a v e r b - c a s e - l i k e prototype to r e s o l v e a sentence having the verb " t o c o n t a i n " . In a previous s e c t i o n i t was pointed out t h a t most case-systems would r e g u i r e another prototype such as "humans d r i n k c o n t a i n e r s " i n order to analyze the above example. I t i s contended here t h a t the use of the same i n f o r m a t i o n , both f o r the purpose of e s t a b l i s h i n g verb-cases, and f o r d e c i s i o n s about f i l l i n g the case r o l e s , provides an i n c r e a s e d semantic p r o d u c t i v i t y , and i n a d d i t i o n , new pi e c e s of i n f o r m a t i o n can be added to the system without c o n t i n u a l l y checking f o r a l l the verb-cases which must be updated to allow f o r t h a t new i n f o r m a t i o n . H i e r a r c h i e s And G e n e r a l i z a t i o n s A Knowledge I d e n t i f i c a t i o n System 32 The nodes of t h e h i e r a r c h i c a l r e l a t i o n network correspond t o semantic c a t e g o r i e s . Each node s p e c i f i e s a s e t of words which are i t s acceptable s u b s t i t u t e s i n a prototype i n v o l v i n g t h a t node. Thus, i n a sense, the set of a l l such h i g h e r - l e v e l nodes i m p l i e s a s e t of p r i m i t i v e s to which a l l of the l e x i c a l elements of the system may be reduced. However, meaningful p r o t o t y p i c a l language uses are represented at any l e v e l of p r i m i t i v i z a t i o n . Not a l l of the nodes i n the h i e r a r c h y need to have a d i r e c t correspondence t o n a t u r a l language elements. For example, " r e s t a u r a n t " , "bar", and " t a v e r n " may be a b s t r a c t e d to PLACES-OF-CONSDMPTION ( c a p i t a l i z e d t o i n d i c a t e t h a t i t i s not a l e x i c a l element). I t i s important to note that t h i s s e t of p r i m i t i v e concepts has not been a r b i t r a r i l y chosen. P r i m i t i v e s are developed i n c o n j u n c t i o n with the formation of a g e n e r a l i z e d p r o t o t y p e which employs the p r i m i t i v e s to r e p r e s e n t the language use of l o w e r - l e v e l l e x i c a l elements which have been assigned to t h a t p r i m i t i v e category. C o l l i n s and Q u i l l i a n (1972:323-4) have suggested t h a t o r g a n i z a t i o n occurs i n memory to permit i n f e r e n c e s i n s t o r i n g and a c c e s s i n g semantic i n f o r m a t i o n . S e v e r a l r e l a t i o n s other than ISA a l s o possess t h i s o r g a n i z a t i o n a l c a p a c i t y , such as PART-OF, CAUSALITY, SIMILARITY, e t c . The system being proposed i n t h i s paper i n c l u d e s only the ISA r e l a t i o n because i t appears t o be an a c t i v e i n g r e d i e n t of semantic o r g a n i z a t i o n over the H i e r a r c h i e s And G e n e r a l i z a t i o n s A Knowledge I d e n t i f i c a t i o n System 33 widest range of con t e x t s . C e r t a i n l y , i f one were t o expect a system to a t t a i n a degree of s i m i l a r i t y to human a b i l i t i e s , then o r g a n i z a t i o n over other r e l a t i o n s would a l s o be necessary. The p r o p o s a l to convert prototypes such as "dog i s animal" i n t o p o i n t e r s i n a h i e r a r c h i c a l s t r u c t u r e does not imply adherence to the view t h a t a l l such ISA r e l a t i o n s are r i g i d l y h i e r a r c h i c a l i n human memory. Evidence now i n d i c a t e s t h a t " b a s i c c a t e g o r i e s " are formed out of f a m i l i a r , easy to grasp concepts and t h a t other concepts are developed with these as t h e i r base, r e g a r d l e s s of t h e i r r e l a t i v e d i r e c t i o n s i n an ISA h i e r a r c h y (Rosch and He r v i s , 1975). For example "dog" would be a b a s i c c ategory, out o f which both " c o l l i e " and "canine" could be formed. A s t r i c t h i e r a r c h y of ISA r e l a t i o n s has been i n c o r p o r a t e d i n t o the system a t t h i s point only because i t p r o v i d e s a uniform o r g a n i z a t i o n a l component. The preceding paragraphs have attempted to demonstrate t h a t the use o f an ISA h i e r a r c h y i n c r e a s e s the extent t o which the o p e r a t i o n s of the system correspond to language-use. There are a l s o computation'al advantages. 1) There has been a r e s u l t a n t r e d u c t i o n i n the number of prototypes to be c o n s i d e r e d . 2) The e x i s t e n c e o f an ISA r e l a t i o n has been s i n g l e d out as the f i r s t c r i t e r i o n to be considered i n t r a c i n g two components. H i e r a r c h i e s And G e n e r a l i z a t i o n s A Knowledge I d e n t i f i c a t i o n System 34 3) The i n t r o d u c t i o n o f ISA r e l a t i o n s has reduced the measure of r e s o l u t i o n from what i t was i n the b a s i c model, f o r the same i n p u t sentence i n v o l v i n g e s s e n t i a l l y the same knowledge. Thus, an i n p u t sentence which i s s i m i l a r to known language experiences i s more meaningful i f those experiences have been g e n e r a l i z e d . This appeals to i n t u i t i o n about language use. 4c. Verb H i e r a r c h i e s and Wore G e n e r a l i z a t i o n What processes c o u l d permit the g e n e r a l i z a t i o n o f s p e c i f i c p r o t o t y p e s i n t o the more ge n e r a l c a s e - l i k e prototypes? The answer to t h i s q u e s t i o n i s not at a l l c l e a r , but some i n d i c a t i o n s can be g i v e n . I f i n p u t s are o f t e n encountered which r e s o l v e e q u a l l y w e l l t o any set of prototypes, then t h a t s e t would have to be c o n s i d e r e d as a candidate f o r g e n e r a l i z a t i o n . The components of these prototypes could be examined to see i f they al r e a d y share some common semantic category. One e s s e n t i a l requirement of these processes i s that g e n e r a l i z a t i o n e i t h e r co-occur with the formation of h i e r a r c h i e s , or be found t o be c o n s i s t e n t with a l r e a d y e x i s t i n g semantic c a t e g o r i e s . For now, however, the system i s c o n s i d e r e d to be s t a t i c i n the sense t h a t new prototypes are not being g e n e r a l i z e d as an ongoing process, but are s p e c i f i e d a t the onset. A dynamic Verb H i e r a r c h i e s and More G e n e r a l i z a t i o n I Knowledge I d e n t i f i c a t i o n System 35 system which c o n t i n u a l l y examined i t s prototypes f o r p o s s i b l e g e n e r a l i z a t i o n s would be l e a r n i n g the semantics of the language. Some i n d i c a t i o n o f the i n t r i c a c i e s of the process o f g e n e r a l i z a t i o n have been provided by Becker (1969). H i s system ( h i s terminology given i n parentheses) d e f i n e s a more r e s t r i c t i v e form of t r a c i n g (analogy mapping) between components (nodes) of prototypes ( k e r n e l s ) . The process of g e n e r a l i z a t i o n ( i n d u c t i o n ) i s demonstrated through the use o f examples t o be deeply i n v o l v e d with the process o f r e s o l u t i o n (forming analogy mappings). Even though the processes of g e n e r a l i z a t i o n are not completely understood, t h e i r e f f e c t s should be c o n s i s t e n t over a l l of the prototypes i n the system. In the pre v i o u s examples, any prototypes being g e n e r a l i z e d have always c o n s i s t e d of s p e c i f i c , l o w - l e v e l knowledge such as: Sam d r i n k s t e a Tom d r i n k s milk ! Humans d r i n k l i q u i d Sam, Tom <human> t e a , milk <liquid> I t should be noted t h a t a prototype which i s already a g e n e r a l i z a t i o n can a l s o take part i n such a formation, as i n : Verb H i e r a r c h i e s and More G e n e r a l i z a t i o n A Knowledge I d e n t i f i c a t i o n System 3 6 humans drink l i g u i d bears d r i n k l i g u i d zebras d r i n k l i g u i d v animals d r i n k l i g u i d humans, bears, ,zebras <animals> Thus knowledge i s r e p r e s e n t e d i n the system at both a b s t r a c t and s p e c i f i c l e v e l s , as w e l l as at any i n t e r m e d i a t e l e v e l . I t i s important to r e a l i z e that the type of knowledge being d i s c u s s e d here i s about the a c c e p t a b i l i t y of language-usages, not r e p r e s e n t a t i o n s of f a c t s about concepts. I f the system has a prototype f o r "monkeys eat bananas", i t does not imply t h a t a l l monkeys eat bananas, or that any s p e c i f i c monkey eats bananas. I t only means that the p a r t i c u l a r combination of semantic concepts i s a c c e p t a b l e , and t h a t i t c o u l d be used to determine the a c c e p t a b i l i t y of other i n p u t s . As was p o i n t e d out i n an e a r l i e r s e c t i o n , the i d e n t i f i c a t i o n of r e l e v a n t prototypes i s only intended as a s t a r t i n g p o i n t i n the s o l u t i o n of any s p e c i f i c l a n g u a g e - r e l a t e d problem. The r e p r e s e n t a t i o n of f a c t s about the " r e a l - w o r l d " world c e r t a i n l y i s necessary i n a complete language understanding system, and would be most l i k e l y connected v i a the more s p e c i f i c p r o t o t y p e s . To t h i s p o i n t , o n l y nouns (subject and object) have been a b l e to take p a r t i n the h i e r a r c h i c a l r e l a t i o n s . Verbs, and i n Verb H i e r a r c h i e s and More G e n e r a l i z a t i o n A Knowledge I d e n t i f i c a t i o n System 37 f a c t any part of speech, can a l s o form h i e r a r c h i e s . For example animals eat food animals d r i n k l i g u i d animals i n g e s t FOOD-SUBSTANCES food, l i q u i d <FOOD-SUBSTANCES> ea t , d r i n k <ingest> what e f f e c t does the e x i s t e n c e of verb h i e r a r c h i e s have upon the process of r e s o l u t i o n ? F i r s t o f a l l , s i n c e each l e x i c a l element can access a l l p r o t otypes i n which i t i s i n v o l v e d , the h i e r a r c h i e s of l e x i c a l elements a l s o r e p r e s e n t a h i e r a r c h y of the p r o t o t y p e s . P r e v i o u s l y i t was mentioned that the verbs of an i n p u t and a r e s o l v e d prototype should match. I f such a r e s o l u t i o n i s not p o s s i b l e , the prototypes of the verb above the i n p u t verb (abstractions) and below i t (instantiations) w i l l be c o n s i d e r e d . For example: George d r i n k s c i d e r George <human> d r i n k <ingest> human <animal> c i d e r <food> animals i n g e s t food Verb H i e r a r c h i e s and More G e n e r a l i z a t i o n A Knowledge I d e n t i f i c a t i o n System 38 An i n t e r e s t i n g g u e s t i o n which a r i s e s out of the proposed system i s whether or not s p e c i f i c i n f o r m a t i o n should be r e t a i n e d once i t has been g e n e r a l i z e d . On the one hand, i f i t i s not r e t a i n e d , the t o t a l amount of data c o u l d be reduced, and as w e l l the number of prototypes being c o n s i d e r e d c o u l d be minimized. On the other hand, i t would be a great advantage to be a b l e to vary the l e v e l of i n s t a n t i a t i o n of a given i n p u t . In e i t h e r c a s e , i t i s d e s i r e a b l e to have verb h i e r a r c h i e s , e i t h e r t o allow more g e n e r a l i z a t i o n , or as a means of s p e c i f y i n g i n s t a n t i a t i o n s and a b s t r a c t i o n s . The process of r e s o l u t i o n could be even more e f f e c t i v e l y c o n t r o l l e d through the use of verb h i e r a r c h i e s . While there may be numerous prototypes using a s p e c i f i c verb, there may be only very few at the upper reaches of the verb h i e r a r c h i e s . Each verb could thus i n d i c a t e some h i g h - l e v e l g e n e r a l i z a t i o n s f o r which r e s o l u t i o n attempts could be made. I f such a r e s o l u t i o n i s p o s s i b l e , then the prototypes which caused the g e n e r a l i z a t i o n c o u l d be c o n s i d e r e d i n search of an i n s t a n t i a t i o n which would provide a b e t t e r measure of r e s o l u t i o n , and so on u n t i l some p o r t i o n of the s p e c i f i c verb's prototypes are being c o n s i d e r e d . For example, the i n p u t "Tom d r i n k s milk" c o u l d be r e s o l v e d f i r s t t o "ANIMATE consume FOOD-SUBSTANCE", which could then be i n s t a n t i a t e d t o "animal i n g e s t food", which would f u r t h e r be Verb H i e r a r c h i e s and More G e n e r a l i z a t i o n A Knowledge I d e n t i f i c a t i o n System 39 i n s t a n t i a t e d to "human d r i n k l i q u i d " , e t c . T h i s p o i n t i s only made here t o i n d i c a t e t h a t the r e t e n t i o n of s p e c i f i c p r o t o types does not n e c e s s a r i l y imply an u n c o n t r o l l a b l e i n c r e a s e i n the time necessary to search through the data. Other mechanisms f o r c o n t r o l l i n g and a c c e s s i n g the prototypes w i l l be d i s c u s s e d i n the s e c t i o n about the implemented model, and more arguments i n f a v o u r of r e t a i n i n g s p e c i f i c prototypes w i l l be presented i n a l a t e r s u b - s e c t i o n . \ Forming Su b - P r o p o s i t i o n s I f the semantic o p e r a t i o n s o u t l i n e d here are t o ever a c t u a l l y f u n c t i o n i n a language a n a l y s i s system, complex and v a r i e d sentence s u r f a c e s t r u c t u r e s must be c o n s i d e r e d . Two p o s s i b l e s o l u t i o n s to t h i s problem are d i s c u s s e d . The f i r s t i n v o l v e s an i n i t i a l p r o c e s s i n g of the i n p u t i n t o a s t r u c t u r e composed of e s s e n t i a l l y s u b j e c t , verb and o b j e c t t r i p l e s . , The second (to be d i s c u s s e d i n the next sub-section) proposes t h a t the form of the prototype be expanded t o i n c l u d e a p r o v i s i o n f o r more complex grammatical forms. Sentences can o f t e n be reduced i n t o a s e r i e s of s h o r t p r o p o s i t i o n s . For example: "The man on the bus with the s i l k hat i s a diplomat." c o u l d be reduced to : Forming S u b - P r o p o s i t i o n s A Knowledge I d e n t i f i c a t i o n System 40 (man be diplomat) (man l o c a t i o n bus) (? has hat) or (? accompanies hat) (hat i s s i l k ) There are some immediate d i f f i c u l t i e s with t h i s approach. L i n e three of the above example show (by the use of a quest i o n mark) t h a t o f t e n the intended r e f e r e n c e s may not be e a s i l y determined, and as w e l l that s e v e r a l a l t e r n a t e p r o p o s i t i o n s can have i d e n t i c a l s u r f a c e forms. I f each p r o p o s i t i o n , with each of i t s p o s s i b l e r e f e r e n c e b i n d i n g s , were s u b j e c t to the process of r e s o l u t i o n d e s c r i b e d e a r l i e r , then only the s e m a n t i c a l l y a c c e p t a b l e ones would have low measures of r e s o l u t i o n . Consider the f o l l o w i n q example of the r e s o l u t i o n of one such reduced i n p u t ("musical instrument" i s a b r e v i a t e d by MI): B i l l p l a y s Tom»s piano B i l l play piano Tom own piano Bill<human> piano<MI> Tom<human> piano<MI> human play MI human own MI Forming S u b - P r o p o s i t i o n s A Knowledge I d e n t i f i c a t i o n System 41 Since the two r e s o l u t i o n s i n the above example are not at a l l independent, there must be some i n t e r a c t i o n between the p r o t o t y p e s used. T h i s i s e s s e n t i a l l y the approach t h a t has been taken by Wilks (1975a); an i n p u t i s segmented, and each segment matched to a "bare-template" (a h i g h l y g e n e r a l i z e d p r o t o t y p e ) , and then " t i e s " are made among the templates., The main problems with h i s approach i s t h a t the i n t e r a c t i n g components of the p r o t o t y p e s are not being c o n s i d e r e d at the time the r e s o l u t i o n s are t a k i n g p l a c e . Such i n f o r m a t i o n could serve to reduce the number of prototypes being c o n s i d e r e d . Many parts of speech seem t o reduce e a s i l y i n t o separate s u b - p r o p o s i t i o n s , such as a d j e c t i v e s , a d j e c t i v e phrases, and r e l a t i v e c l a u s e s . Other p a r t s of speech do not. Adverbs and a d v e r b i a l phrases are p a r t i c u l a r l y d i f f i c u l t ; f o r example, how c o u l d one r e p r e s e n t the p r e p o s i t i o n a l phrase i n : "He b u t t e r e d h i s t o a s t with h i s k n i f e . " The complex problems of r e d u c i n g an i n p u t to a s t r u c t u r e of s h o r t p r o p o s i t i o n s can best be seen i n the work of Anderson and Bower (1972). Another problem with such a method of reducing the input i s the f a c t that a s i n g l e p r o p o s i t i o n c o u l d be represented i n s e v e r a l d i f f e r e n t ways i n a sentence. For example "humans own p i a n o s " c o u l d be represented as: Forming S u b - P r o p o s i t i o n s A Knowledge I d e n t i f i c a t i o n System 42 human's piano piano of human piano i s human's piano belongs to human Co n s i d e r the p o s s i b i l i t y t h a t the prototypes maintain i n f o r m a t i o n about t h e i r p o s s i b l e s u r f a c e s t r u c t u r e m a n i f e s t a t i o n s . T h i s i n f o r m a t i o n could be a s s o c i a t e d with the hi g h e s t p o s s i b l e l e v e l s o f g e n e r a l i z a t i o n , to av o i d redundancy, but there would s t i l l be a problem. The only c o n t r o l on the proc e s s of s e a r c h i n g through the prototypes has been t h a t the verb was a l r e a d y known, which would not be the case i f an attempt was made to use the prototypes to determine what b a s i c p r o p o s i t i o n s were intended. On the other hand i t i s c l e a r t h a t i n most cases the grammatical r e l a t i o n s of the s u r f a c e s t r u c t u r e are not enough to suggest the a p p r o p r i a t e p r o p o s i t i o n . . An i n i t i a l p r e - p r o c e s s i n g i n t o p r o p o s i t i o n s i s an appealing i d e a from w i t h i n the framework of the system being developed here, but a c l o s e r examination r e v e a l s t h a t i t s problems are too numerous and d i f f i c u l t . 4e. Expanding the Content of the jProtg.ty.pe More complex grammatical forms can be analyzed i f the cont e n t of the prototype i s expanded. For example, a prototype might be as f o l l o w s : Expanding the Content of the Prototype A Knowledge I d e n t i f i c a t i o n System 43 humans play piano I 1 with hands The attachment t o the verb " p l a y " i s i n d i c a t e d by the v e r t i c a l l i n e . T h i s e x t r a i n f o r m a t i o n "with hands" c o u l d be thought of as the s p e c i f i c a t i o n of an a d d i t i o n a l case r o l e . The usual case n o t a t i o n has not been i n c l u d e d i n order t o s i m p l i f y the examples and to avoid a commitment t o any p a r t i c u l a r system of case names. Any i n p u t sentence, as w e l l as any pro t o t y p e , has a basic t r i p l e , which i s simply the s u b j e c t , verb and o b j e c t t r i p l e (the o b j e c t may be omitted) which were mentioned e a r l i e r . U n l i k e s u b - p r o p o s i t i o n s , i t i s a r e l a t i v e l y simple task to i d e n t i f y the b a s i c t r i p l e of an i n p u t sentence, and t o match i t t o the ba s i c t r i p l e of some prototype. Then an attempt c o u l d be made to i d e n t i f y the remaining s u r f a c e s t r u c t u r e with case p o s s i b i l i t i e s , such as i n t h i s example: Dick played the keyboard with h i s hands Dick<human> keyboard part-of piano humans pl a y piano with hands Expanding the Content of the Prototype A Knowledge I d e n t i f i c a t i o n System 44 Should some aspect of the s u r f a c e s t r u c t u r e not be i d e n t i f i e d i n t h i s way, then the prototypes o f the words i n v o l v e d i n t h a t s t r u c t u r e (and t h e i r a b s t r a c t i o n s ) could be examined to l o c a t e one which might serve as an a d d i t i o n a l p r o p o s i t i o n . For example, the sentence "The man with the g u i t a r sang the song" might r e s o l v e (as a b a s i c t r i p l e ) to "human s i n g song", which does not have any case r o l e which s p e c i f i e s what the "human" might "have". Thus the prototypes i n v o l v i n g " g u i t a r " and "musical instrument" would be checked to see i f any of them i n v o l v e "man" (or one o f i t s a b s t a c t i o n s ) and, "human p l a y musical instrument" would be s e l e c t e d because i t s components t r a c e well to the b a s i c t r i p l e . Of course, the p r o p o s i t i o n would have to i n v o l v e the part o f speech found i n the s u r f a c e s t r u c t u r e . When prototypes with e x t r a case r o l e s are g e n e r a l i z e d , t h e i r a d d i t i o n a l components may or may not be i n c l u d e d . Thus, f o r example, the two prototypes given below may be g e n e r a l i z e d t o not i n c l u d e t h e i r components. Expanding the Content of the Prototype A Knowledge I d e n t i f i c a t i o n System 45 humans play piano I I with hands humans play harmonica I ! with mouth I I V humans play musical instruments harmonica, piano <musical instrument> In t h i s case, the o r i g i n a l s p e c i f i c p r o t o types must be r e t a i n e d i n the system, or e l s e important i n f o r m a t i o n w i l l be l o s t . Even i n the event t h a t the a d d i t i o n a l components are themselves g e n e r a l i z e d , the more s p e c i f i c prototypes should s t i l l be r e t a i n e d (as w i l l be shown l a t e r on). humans p l a y piano \ with hands humans play harmonica I with mouth I I V humans play musical instruments I with BODY-PARTS harmonica, piano <musical instrument> mouth, hands <BODY-PARTS> Expanding the Content of the Prototype A Knowledge I d e n t i f i c a t i o n System 46 Most i n s t a n c e s o f g e n e r a l i z a t i o n w i l l r e t a i n a l l of the i n v o l v e d prototypes. T h i s c o u l d be a burden on the storage requirements of the system, but, i t does a l l o w an important f l e x i b i l i t y i n p r o c e s s i n g . Consider a system with the f o l l o w i n g p r o t o t y p e s : waiters serve FOOD-SOBSTANCES I i n PLACES-OF-CONSUMPTION wai t e r s serve beer waiters serve d r i n k s I 1 i n t a v e r n s i n bars beer, d r i n k s <FOOD-SUBSTANCES> bars, t a v e r n s <PLACES-OF-CONSUMPTION> The f i r s t prototype has been g e n e r a l i z e d from the oth e r two. In the a n a l y s i s of the sentence: "The waiter s e t the g l a s s i n f r o n t of Tom." the g e n e r a l i z e d prototype would be s u f f i c i e n t to allow r e s o l u t i o n , and i n f a c t there would be no way of d e c i d i n g which o f the other two was intended. However i f the sentence was: "The waiter placed the g l a s s i n f r o n t o f Tom j u s t as the domino game s t a r t e d . " then a connection may not be made between "dominos" and any component of the t a r g e t prototype, and so i t would be necessary t o c o n s i d e r the i n s t a n t i a t i o n s of the t a r g e t . The s p e c i f i c p l a c e " t a v e r n " would be chosen beacuse the connection "people p l a y dominos i n t a v e r n s " can be made. Expanding the Content of the Prototype A Knowledge I d e n t i f i c a t i o n System 47 B e c a l l t h a t the method of s e a r c h i n g through the prototypes was to s t a r t at those which were the most g e n e r a l i z e d . These more ge n e r a l prototypes could provide t e n t a t i v e r e s o l u t i o n s , which c o u l d be i n s t a n t i a t e d when the requirement a r i s e s , e i t h e r through an a d d i t i o n a l component of the sentence being encountered, or through a subsequent sentence, or even through some background f e a t u r e of the language use s i t u a t i o n , such as s o c i a l p o s i t i o n s of speakers, e t c . This aspect o f the use of the model i s extremely s p e c u l a t i v e , and i s only mentioned to i n d i c a t e a p o s s i b l e d i r e c t i o n f o r f u r t h e r study. Expanding the Content of the Prototype Some Fu r t h e r Issues 48 5. Some F u r t h e r Issues T h i s s e c t i o n i s concerned with some i s s u e s which a r i s e out o f the theory which has been d i s c u s s e d thus f a r . I t i s only i n t e n d e d to provide some p o s s i b l e d i r e c t i o n s f o r f u t u r e study by p o i n t i n g out the problems and p o s s i b i l i t i e s . 5a. T r a c i n g In the previous s e c t i o n s i t has been noted t h a t the reason why l e x i c a l elements may be t r a c e d i s due t o e i t h e r a h i e r a r c h i c a l r e l a t i o n or the e x i s t e n c e of some prototype which i n v o l v e s both of them. I t appears t h a t the r e l a t i o n s can be used independently of the a c t u a l words themselves. That i s , i f any prototype "A B C" e x i s t s , then any l e x i c a l element beneath "A" i n the h i e r a r c h y can be used i n i t s p l a c e . T h i s i s not, of course, t r u e f o r the more complicated type of t r a c i n g . The sentence: "The c i t y l o v e s parades." i s a c c e p t a b l e because of the prototype " c i t i e s c o n t a i n people", and thus a r e s o l u t i o n can take place t o the prototype "people l o v e parades". On the other hand, the sentence: "the doorknob owns the piano" i s not a c c e p t a b l e j u s t because "humans turn doorknobs". The T r a c i n g Some Further Issues 49 p o i n t i s that the use of a prototype independent of any s p e c i f i c knowledge about the words or concepts i n v o l v e d may permit the acceptance of sentences which are not o r d i n a r i l y meaningful. The frequency with which meaningful sentences might be m i s i n t e r p r e t e d may not, however, j u s t i f y c o n s t a n t l y r e s o r t i n g to s p e c i f i c knowledge ( r e c a l l t h a t the best measure of r e s o l u t i o n i s always s e l e c t e d ) . Consider f o r example a sentence such as: "Tom ate the cupboard." I t i s necessary to t r a c e "cupboard" to "food" i n order t o r e s o l v e "humans eat food". I t i s d i f f i c u l t t o imagine another c h a i n connecting "cupboard" to some other o b j e c t which humans eat , which i s s h o r t e r than would be obtained with "cupboards c o n t a i n food". So w i t h i n the context of a system which expects t o be presented with meaningful sentences o n l y , and which i s committed to make "the best p o s s i b l e " sense of any i n p u t , perhaps t h i s type of uniform t r a c i n g i s reasonable as a f i r s t s t e p i n the a n a l y s i s of an i n p u t . I t was hoped t h a t the p o s i t i o n a l r e l a t i o n s among the components of a prototype would provide some c l u e as t o t h e i r u t i l i t y i n t r a c i n g . But co n s i d e r the two i n p u t s : "Tom empties h i s c o f f e e . " "Tom drank the cup." Both of these sentences are t r a c e d through the prototype "cups T r a c i n g Some Further Issues 50 c o n t a i n l i q u i d " . In one case the s u b j e c t i s being t r a c e d , and i n the other the o b j e c t . T h i s seems t o r u l e out any dependence on p o s i t i o n a l r e l a t i o n s . another problem concerning t r a c i n g i n v o l v e s the use of the t a r g e t prototype as a t r a c i n g prototype. Suppose an input X i s being r e s o l v e d t o a prototype Y then should Y be used t o t r a c e any of the components? There are examples i n which doing so w i l l produce inadequate r e s u l t s , such as: Tom pl a y s Beethoven (input) Tom<human> Beethoven<human> human play MI human play MI (prototype) In t h i s example, the prototype "humans pl a y MI" has been used both as the t a r g e t and as pa r t of the t r a c i n g o f "Beethoven" t o "MI". However, the f a c t s t h a t "Beethoven wrote music" and "MI emit music" would have been a b e t t e r c h o i c e of t r a c i n g p r o t o t y p e s (see example 5 i n s e c t i o n 6a). The use of the t a r g e t prototype i n t r a c i n g leads to a general problem: i f a sentence "a B C" r e s o l v e s to a prototype "X Y Z", then so would "A B A". T r a c i n g Some Furth e r Issues 51 A B C T h i s i s c l e a r l y u n s a t i s f a c t o r y , but t o i n h i b i t the use of the t a r g e t prototype i n t r a c i n g would d i s a l l o w the f o l l o w i n g : John p l a y s Tom (input) John<human> Tom<human> human play games human play games (prototype) which i s , of course, a reasonable r e s o l u t i o n , which capt u r e s the i d e a of the r e l a t i o n between the v a r i a t i o n s i n t h i s sense o f the verb " t o p l a y " . The expanded content form of prototypes can a v o i d the problem f o r t h i s p a r t i c u l a r i n s t a n c e : T r a c i n g Some Further Issues 52 John p l a y s Tom (input) John<human> Tom<human> human play games with humans but t h i s may not be a gen e r a l s o l u t i o n to the problem. 5b. D i r e c t e d I n s t a n t i a t i o n P r e v i o u s l y i t was i n d i c a t e d t h a t a g e n e r a l i z a t i o n need not r e t a i n a l l of the c a s e - r o l e s that appear i n the more s p e c i f i c p r o t o t y p e s from which i t was c r e a t e d . I t has a l s o been mentioned t h a t i n the event t h a t i n s u f f i c i e n t c a s e - r o l e s e x i s t i n a prototype which i s the t a r g e t of the r e s o l u t i o n , an i n s t a n t i a t i o n or a b s t r a c t i o n could be attempted. c o n s i d e r the example: D i r e c t e d I n s t a n t i a t i o n Some Farth e r Issues 53 B i l l plays Beethoven with h i s f e e t B i l l <human> humans p i Beethoven Beethoven Y MI piano <MI> . humans play piano f e e t are s i m i l a r wrote music played piano to hands MI emits music with hands The c o n t r o l mechanisms which would permit t h i s r e s o l u t i o n are beyond what has been p r e v i o u s l y d i s c u s s e d . The "humans play m u s i c a l i n s t r u m e n t s " prototype (MI i s used to a b b r e v i a t e musical instrument) was being r e s o l v e d g u i t e n i c e l y i n a way s i m i l a r to what has been shown b e f o r e , but the p r e p o s i t i o n a l phrase could not ( i t i s assumed) be connected i n t h i s r e s o l u t i o n . Thus an i n s t a n t i a t i o n of the prototype was attempted (to one of the more s p e c i f i c p r o t o types used to form the more g e n e r a l one). The e x i s t e n c e of "piano" i n the prototype "Beethoven played piano" D i r e c t e d I n s t a n t i a t i o n Some Furth e r Issues 54 has a s s i s t e d i n t h i s i n s t a n t i a t i o n . In other words, wherever p o s s i b l e , an i n s t a n t i a t i o n should be c o n s i s t e n t with the r e s t of the input (or perhaps c o n t e x t ) , F i n a l l y , s i n c e the lower l e v e l p r ototype has the a d d i t i o n a l content (with hands) then the p r e p o s i t i o n a l phrase i n the i n p u t can be t r a c e d to i t . Note a l s o , t h a t the prototype used t o t r a c e these components could i t s e l f have been a ch a i n of t r a c e d p r o t o t y p e s . In g e n e r a l , i f d i f f e r e n t knowledge e x i s t e d , a d i f f e r e n t r e s o l u t i o n would be formed. 5c, Metaphor Any use of a prototype i n t r a c i n g the components of a r e s o l u t i o n r e p r e s e n t s a metaphorical use of n a t u r a l language, C o l l i n s and Q u i l l i a n (1972:347) have pointed out the continuum r a n g i n g from s t r a i g h t forward meaningful sentences to metaphor t o anomaly. The system being proposed i n t h i s paper, i n c l u d e s the a b i l i t y to understand two types of f i g u r a t i v e language. Metonymy i s the r e f e r e n c e t o the accompaniament of a t h i n g i n s t e a d o f the t h i n g i t s e l f , such as "The pen i s m i g h t i e r than the sword." and "The t a b l e looked d e l i c i o u s . " T h i s i s , of course, o n l y the use of another prototype t o t r a c e the unusual component. In the event t h a t the PART-OF r e l a t i o n i s being used, then t h i s type o f metaphor i s termed "synecdoche" Metaphor Some F u r t h e r Issues 55 (see W i l l i s , 1973) as i n : " A l l hands on deck." and "My wheels are i n the garage." W i t h i n the present system, t h e r e i s no p r o v i s i o n t o d i s t i n g u i s h t h e PABT-OF r e l a t i o n from the o t h e r s . There i s , however, a s p e c i a l emphasis on the ISA r e l a t i o n . The use of the h i e r a r c h i c a l r e l a t i o n s i s more l i m i t e d than the use of the ether p r o t o t y p e s ; so, to allow a c o n s i s t e n t acceptance of metaphoric language use, the r u l e s f o r t r a c i n g would have t o be a l t e r e d . The sentence "The sergeant barked h i s o r d e r s . " i s a c c e p t a b l e because of the s i m i l a r i t y between "dog" and "human" (which may be more exaggerated i n the case of the sergeant who i s being r e f e r e n c e d ) . Part of t h i s s i m i l a r i t y i s captured by the f a c t t h a t both "human" and "dog" l i e beneath "animal" (or perhaps "mammal") w i t h i n the h i e r a r c h i c a l network, The present r u l e s do not allow such a t r a v e r s a l of the network, but t h i s c o u l d be i n c o r p o r a t e d by a l l o w i n g S to t r a c e to T i f some element B were d i r e c t l y above each o f them, or, more g e n e r a l l y , i f some element B i s d i r e c t l y above T and anywhere above S i To extend the r u l e s i n t h i s way would a l s o permit another, more n a t u r a l , language use. Suppose the system has a pr o t o t y p e "Tom pla y s b a s e b a l l " but does not have the g e n e r a l i z a t i o n t h a t "humans p l a y b a s e b a l l " . However, the system Metaphor Some F u r t h e r Issues 56 does r e a l i z e (through some other g e n e r a l i z a t i o n ) t h a t both "Tom" and " B i l l " are humans. Then t h i s new type of t r a c i n g could be used to l o c a t e "Tom p l a y s b a s e b a l l " as a r e s o l u t i o n f o r " B i l l p l a y s b a s e b a l l " . I t i s a l s o p o s s i b l e t h a t h i e r a r c h i c a l r e l a t i o n s are not formed among elements which are q u i t e s i m i l a r . I f the number of s i m i l a r i t i e s (in terms of a c c e p t a b l e uses of the words) become g r e a t , perhaps they may e v e n t u a l l y be subsumed by a s i n g l e concept i n the h i e r a r c h i c a l r e l a t i o n network; however, i n the meantime, i t i s d e s i r a b l e to be able to use t h e i r s i m i l a r i t y to r e s o l v e i n p u t s . T h i s c o u l d be accomplished by extending the r u l e s to i n c l u d e : S t r a c e s t o T i f they share some number of p r e d i c a t e s (verb-object p a r t s of the p r o t o t y p e s ) . I n c l u d i n g a l l methods thus f a r , t r a c i n g can be d e f i n e d as: L e t t i n g A<B i n d i c a t e that B i s d i r e c t l y above A i n the ISA h i e r a r c h y , then S-->T[p] with Z'=Z-p i f f 1) S=T or 2) S<T> or 3) OR) S<R> S T<R or (3q#rez.,-p) q=SXY & r=TX Y or 5) Q q e Z ' ) S6q & (3R6q) R#S S R—>T[p] with Z»-q Metaphor Some Further Issues 57 T h i s means th a t component S t r a c e s t o component T i f one of the f o l l o w i n g : 1) they are the same l e x i c a l element 2) T i s an a b s t r a c t i o n of S 3) some R i s an a b s t r a c t i o n of both S and T (must be an immediate a b s t r a c t i o n of T) 4) S and T share p r e d i c a t e s (a p r e d i c a t e being a ve r b - o b j e c t p a i r ) 5) some S i s found i n a prototype i n v o l v i n g S, and R t r a c e s to T 5d. Learning In an e a r l i e r s e c t i o n i t was noted t h a t the i d e n t i f i c a t i o n o f p i e c e s of knowledge (prototypes) which were used i n r e s o l u t i o n was to be thought of as a f i r s t process i n v o l v e d i n any language use s i t u a t i o n . While i t i s not w i t h i n the scope of t h i s paper to c o n s i d e r the subsequent process which would use the i d e n t i f i e d i n f o r m a t i o n to access other forms of knowledge to perform any a r b i t r a r y language a p p l i c a t i o n , t h e r e are i n d i c a t i o n s t h a t language l e a r n i n g i s p o s s i b l e w i t h i n t h i s model. These i n d i c a t i o n s take the form of three o b s e r v a t i o n s : Learning ome Further Issues 58 1 ) The model has been designed t o maintain access t o the i n p u t sentence i n a form which i s s i m i l a r t o the form of the knowledge of the system. T h i s could f a c i l i t a t e the i n c o r p o r a t i o n of knowledge, 2) The o r g a n i z a t i o n of knowledge i n the model i s based upon i n d i v i d u a l fragments of i n f o r m a t i o n , and the e f f e c t of the i n c o r p o r a t i o n of new knowledge can be a n t i c i p a t e d , even though the processes of a s s i m i l a t i o n are yet to be designed, 3) The e x i s t e n c e of the r e s o l u t i o n measure p r o v i d e s the p o s s i b i l i t y f o r c r i t e r i a f o r the i n c o r p o r a t i o n of new knowledge. The knowledge i n a sentence would not be i n c o r p o r a t e d i f i t matched d i r e c t l y (low r e s o l u t i o n measure) to a prototype, i f i t were i n f a c t a copy of a p r e v i o u s experience. On the o t h e r hand, i f a long i n v o l v e d r e s o l u t i o n i s necessary, t h a t i s , i f the sentence i s j u s t b a r e l y meaningful (high r e s o l u t i o n measure) then i n c o r p o r a t i o n does not seem reasonable. So perhaps only those i n p u t s whose measure f a l l s i n some middle ground should be i n c o r p o r a t e d . Learning an Implemented Model 59 6. an Implemented Model A s e r i e s of LISP programs has been w r i t t e n t o i n v e s t i g a t e the f e a s i b i l i t y o f the system as d e s c r i b e d , and to c l a r i f y i t s o p e r a t i o n through the use of examples. About 50 l e x i c a l elements, maintaining v a r i o u s h i e r a r c h i c a l r e l a t i o n s to one another, and 29 prototypes a t v a r i o u s l e v e l s of g e n e r a l i z a t i o n have been i n c l u d e d . Each l e x i c a l element accesses (under the pro p e r t y UP) those words which appear d i r e c t l y above i t i n the ISA h i e r a r c h y . In a d d i t i o n , each word acce s s e s (under property PROTOTYPES) a l i s t o f the prototypes i n which i t i s found. These l i s t s are ordered (by the way they were input) so t h a t the most ge n e r a l p r o t o t y p e s appear f i r s t . Each such prototype i s a s u b j e c t , verb, o b j e c t t r i p l e , as are the i n p u t s . For each i n p u t , attempts are made to r e s o l v e i t a g a i n s t a l l p r o t o t y p e s accessed by any verb appearing above the input verb i n the network, o r , of course, accessed by the in p u t verb i t s e l f . The process of r e s o l u t i o n i s b a s i c a l l y the same as d e s c r i b e d at the end of s e c t i o n f i v e . Each component of the i n p u t i s t r a c e d to the prototype components, t r y i n g each of the c o n d i t i o n s f o r t r a c i n g i n the given order. As soon as one t r a c e i s found, then no other i s attempted ( f o r t h i s reason the " s t r o n g e s t " t r a c i n g methods are attempted f i r s t ) . An Implemented Model An Implemented Model 60 Each such p o s s i b l e r e s o l u t i o n i s measured by a procedure which a s s i g n s values to each type of t r a c e , and sums them. The lowest measure i s the best. The outermost r o u t i n e accepts an i n p u t as i t s argument, and d i s p l a y s a l l r e s o l v e d p r o t o t y p e s , a l o n g with t h e i r o v e r a l l measure. The t r a c e used f o r each component i s a l s o d i s p l a y e d i n a readable form. 6a. Sample Runs The f o l l o w i n g i s a d i s p l a y of the words f o r which the system has h i e r a r c h i c a l r e l a t i o n s h i p s , d i s p l a y e d along with each o f t h e i r ISA h i e r a r c h y paths. Note t h a t any l e x i c a l element may be subsumed by s e v e r a l d i f f e r e n t c a t e g o r i e s . * (ALL-HIER) > > > > > > > > > > > > > > > > > > > > > > > > > > ALL HIERARCHIES ABSORB (CONSUME) ANIMAL (ANIMATE) APPLE (FOOD SOLID SUBSTANCE) APPLE (FOOD FOOD-SUBST) BEETHOVEN (HUMAN ANIMAL ANIMATE) BILL (HUMAN ANIMAL ANIMATE) BISCUIT (FOOD SOLID SUBSTANCE) BISCUIT (FOOD FOOD-SUBST) BLUES (MUSIC) CARROT (PLANT ANIMATE) CaBROT (SOLID SUBSTANCE) CARROT (FOOD SOLID SUBSTANCE) CARROT (FOOD FOOD-SUBST) CEREAL (FOOD SOLID SUBSTANCE) CEREAL (FOOD FOOD-SUBST) CLASSICAL (MUSIC) COFFEE (FOOD SOLID SUBSTANCE) COFFEE (FOOD FOOD-SUBST) COFFEE (LIQUID SUBSTANCE) COLLIE (DOG ANIMAL ANIMATE) CUP (CONTAINER) DOG (ANIMAL ANIMATE) Sample Runs An Implemented Model 61 > DRINK (INGEST CONSUME) > EAT (INGEST CONSUME) > FOOD (SOLID SUBSTANCE) > FOOD (FOOD-SUBST) > GUITAR (INSTRUMENT) > HUMAN (ANIMAL ANIMATE) > INGEST (CONSUME) > JAR (CONTAINER) • > LASSIE (COLLIE DOG ANIMAL ANIMATE) > LIQUID (SUBSTANCE) > MILK (FOOD SOLID SUBSTANCE) > MILK (FOOD FOOD-SUBST) > MILK (LIQUID SUBSTANCE) > PIANO (INSTRUMENT) > PLANT (ANIMATE) > PLANT-FOOD (FOOD-SUBST) > SAM (HUMAN ANIMAL ANIMATE) > SERGEANT (HUMAN ANIMAL ANIMATE) > SOLID (SUBSTANCE) > SUNLIGHT (PLANT-FOOD FOOD-SUBST) > TEA (FOOD SOLID SUBSTANCE) > TEA (FOOD FOOD-SUBST) > TEA (LIQUID SUBSTANCE) > TINY (COLLIE DOG ANIMAL ANIMATE) > TOAST (FOOD SOLID SUBSTANCE) > TOAST (FOOD FOOD-SUBST) > TOM (HUMAN ANIMAL ANIMATE) > WATER (PLANT-FOOD FOOD-SUBST) > WATER (FOOD SOLID SUBSTANCE) > WATER (FOOD FOOD-SUBST) > WATER (LIQUID SUBSTANCE) The f o l l o w i n g i s a l i s t of the prototypes contained i n the system. * (ALL-PROTOS) > > PROTOTYPES KNOWN: > > ((HUMAN PLAY INSTRUMENT)) > ( (BILL PLAY PIANO)) > ( (TOM PLAY GUITAR) ) > ((BEETHOVEN WRITE MUSIC)) > ((INSTRUMENT EMIT MUSIC)) > ; ((HUMAN EMPTY CONTAINER)) > ((CONTAINER CONTAIN SUBSTANCE)) > ((JAR CONTAIN SOLID)) > ( (CUP CONTAIN LIQUID) ) > ( (ANIMATE CONSUME FOOD-SUBST)) Sample Runs An Implemented Model 62 > ((PLANT ABSORB PLANT-FOOD)) > ((PLANT ABSORB WATER)) > ((PLANT ABSORB SUNSHINE)) > ((ANIMAL INGEST FOOD)) > ( (DOG EAT FOOD)) > ((DOG DRINK LIQUID)) > { (HUMAN EAT FOOD)) > {(HUMAN DRINK LIQUID)) > ( (TINY EAT CEREAL)) > ( (DOG BARK NULL)) > ((TINY HAS FLEAS) ) > ( (LASSIE BURY BONE) ) > ((LASSIE EAT BISCUIT)) > ( (TINY DRINK MILK) ) > { (LASSIE DRINK WATER) ) > ( (BILL EAT TOAST)) > ((TOM EAT APPLE)) > ((BILL DRINK COFFEE)) > ( (SAM DRINK MILK) ) example 1 * (SHOW ((TOM DRINK MILK))) > > > ANIMATE CONSUME FOOD-SUBST MEASURE= 7 > TOM IS A ANIMATE > DRINK IS A CONSUME > MILK IS A FOOD-SUBST > > ANIMAL INGEST FOOD MEASURE= 4 > TOM IS A ANIMAL > DRINK IS A INGEST > MILK IS A FOOD > > DOG DRINK LIQUID MEASURE= 9 > TOM AND DOG ARE BOTH (ANIMAL) > MILK IS A LIQUID > > HUMAN DRINK LIQUID MEASURE= 2 > TOM IS A HUMAN > MILK IS A LIQUID > > BILL DRINK COFFEE MEASURE^ 14 > TOM AND BILL ARE BOTH (HUMAN) > MILK AND COFFEE ARE BOTH (FOOD LIQUID) > > SAM DRINK MILK MEASURE= 8 Sample Runs An Implemented Model 63 > TOM AND SAM ARE BOTH (HUMAN) > Example one shows a s e r i e s of r e s o l u t i o n s f o r "Tom dr i n k milk". Notice that f o r each component of the i n p u t which i s d i f f e r e n t from the component of the prototype, a l i n e appears beneath the prototype which e x p l a i n s the t r a c e of t h a t component. The prototype "human dr i n k l i q u i d " has the lowest measure, and would thus be s e l e c t e d as the b a s i c knowledqe i d e n t i f i e d f o r the i n p u t , along with . the t r a c i n g i n f o r m a t i o n . Two a b s t r a c t i o n s were a l s o r e s o l v e d : "animate consume FOOD-SUBST" and "animal i n q e s t food". These a t t a i n e d worse measures of r e s o l u t i o n because of the d i s t a n c e s i n the h i e r a r c h y network between the components t r a c e d . Two o t h e r s : " B i l l d r i nk c o f f e e " and "Sam dr i n k m i l k " are chosen because t h e i r components share h i e r a r c h i c a l i n c l u s i o n s . One " f a l s e metaphor" was a l s o p i c k e d out i n "doq d r i n k l i q u i d " . T h i s example r e p r e s e n t s the use of prototype r e s o l u t i o n i n a very n a t u r a l language-use s i t u a t i o n i n which g e n e r a l i z a t i o n s a l r e a d y e x i s t f o r the i n p u t , hence the very low measures. I f the system had the prototype f o r "Tom d r i n k milk", then , o b v i o u s l y , i t would have been chosen with a measure of zero. Sample Runs An Implemented Model 6a example 2 * (SHOW ((LASSIE EAT JAR))) > > > ANIMAL INGEST FOOD MEASURE* 15 > LASSIE IS A ANIMAL > EAT IS A INGEST > JAR AND FOOD APPEAR IN ( (JAR CONTAIN SOLID)) > FOOD IS A SOLID > > > DOG EAT FOOD MEASURE* 13 >, LASSIE IS A DOG > JAR AND FOOD APPEAR IN ( (JAR CONTAIN SOLID)) > FOOD IS A SOLID > > > HUMAN EAT FOOD MEASURE* 19 > LASSIE AND HUMAN ARE BOTH (ANIMAL) > JAR AND FOOD APPEAR IN ( (JAR CONTAIN SOLID)) > FOOD IS A SOLID > > > TINY EAT CEREAL MEASURE* 20 > LASSIE AND TINY ARE BOTH (COLLIE) > JAR AND CEREAL APPEAR IN ( (JAR CONTAIN SOLID)) > CEREAL IS A SOLID > > > LASSIE EAT BISCUIT MEASURE* 12 > JAR AND BISCUIT APPEAR IN ((JAR CONTAIN SOLID)) > BISCUIT IS A SOLID > > * Example two demonstrates the r e s o l u t i o n f o r an in p u t f o r which no g e n e r a l i z a t i o n e x i s t s . The prototypes "dog eat food" and " L a s s i e eat b i s c u i t " are s e l e c t e d as the best, and the l a t t e r has a s l i g h t p r e f e r e n c e because i t i s more s p e c i f i c . In both examples, " j a r c o n t a i n s o l i d " has been used to t r a c e " j a r " Sample Runs An I m p l e m e n t e d M o d e l 65 t o " f o o d " . exajnjDle * (SHOW ( (SAM EMPTY COFFEE) ) ) > > > HUMAN EMPTY CONTAINER MEASURE* 13 > SAM IS A HUMAN > COFFEE AND CONTAINER APPEAR IN ( (CONTAINER CONTAIN SUBSTANCE > COFFEE IS A SUBSTANCE > * T h i s next example shows t h a t the p r o t o t y p e " c o n t a i n e r c o n t a i n s u b s t a n c e " can be used t o t r a c e components i n a d i f f e r e n t s i t u a t i o n . In t h i s c a s e , t h e system "knows" t h a t numans empty c o n t a i n e r s , and has used the o t h e r i n f o r m a t i o n t o t r a c e " c o n t a i n e r s " to " l i q u i d " . example 4 * (SHOW ((CUP CONTAIN TEA))) > > > CONTAINER CONTAIN SUBSTANCE MEASURE* 3 > CUP IS A CONTAINER > TEA IS A SUBSTANCE > > JAR CONTAIN SOLID MEASURE* 10 > CUP AND JAR ARE BOTH (CONTAINER) > TEA IS A SOLID > > CUP CONTAIN LIQUID MEASURE* 1 > TEA IS A LIQUID > * Sample Runs An Implemented Model 66 Of c o u r s e , the same piece of i n f o r m a t i o n can be used as the t a r g e t prototype i n understanding the more s t r a i g h t - f o r w a r d i n p u t given i n example f o u r . example 5 * (SHOW ((BILL PLAY BEETHOVEN) ) ) > > > HUMAN PLAY INSTRUMENT MEASURE= 21 > BILL IS A HUMAN > BEETHOVEN AND INSTRUMENT APPEAR IN ( (INSTRUMENT EMIT MUSIC) > BEETHOVEN AND MUSIC APPEAR IN ((BEETHOVEN WRITE MUSIC)) > > > BILL PLAY PIANO MEASURE= 21 > BEETHOVEN AND PIANO APPEAR IN ( (INSTRUMENT EMIT MUSIC)) > PIANO IS A INSTRUMENT > BEETHOVEN AND MUSIC APPEAR IN ((BEETHOVEN WRITE MUSIC)) > > > TOM PLAY GUITAR MEASURE= 29 > BILL AND TOM ARE BOTH (HUMAN) > BEETHOVEN AND GUITAR APPEAR IN ((INSTRUMENT EMIT MUSIC)) > GUITAR IS A INSTRUMENT > BEETHOVEN AND MUSIC APPEAR IN ((BEETHOVEN WRITE MUSIC)) > > In s e c t i o n f i v e we saw that the d e f i n i t i o n of t r a c i n g i s r e c u r s i v e . Example f i v e p rovides a case i n which two concepts "Beethoven" and "instrument" are t r a c e d through two pr o t o t y p e s : "instrument emit music" and" Beethoven write music". Even the Sample Runs An Implemented Model 67 best measure f o r any such r e s o l u t i o n i s high. Th i s r e f l e c t s the f a c t t h at i t s meaning i s more obscure than that of " B i l l play piano". example 6 K * (SHOW ( (LASSIE HAS FLEAS))) > > > TINY HAS FLEAS MEASURE* 8 > LASSIE AND TINY ARE BOTH (COLLIE) > * * Example s i x demonstrates one of the ideas d i s c u s s e d c o n c e r n i n g metaphor: t h a t . elements which are subsumed by a s i n g l e element may be t r a c e d t o one another. Example seven shows the s i t u a t i o n given i n s e c t i o n f i v e i n which a more t r u l y metaphoric use i s made of t h i s a b i l i t y . example 7 * (SHOW ( (SERGEANT BARK NULL))) > > > DOG BARK NULL MEASURE* 8 > SERGEANT AND DOG ARE BOTH (ANIMAL) > * An a b i l i t y not p r e v i o u s l y mentioned i s i l l u s t r a t e d i n Sample Runs An Implemented Model 68 example e i g h t : t h a t a g e n e r a l i z a t i o n of a concept may be t r a c e d t o the s p e c i f i c one (with a high e r r e s o l u t i o n measure than v i c e - v e r s a ) . example 8 * (SHOW ((DOG BURY BONE))) > > > LASSIE BURY BONE MEASDEE= 5 > DOG IS A GENERALIZATION OF LASSIE > 6 b. D i s c u s s i o n The goal of producing t h i s implementation was t o provide a b e t t e r d e s c r i p t i o n of the processes i n v o l v e d . I t has been the o v e r a l l i n t e n t of t h i s paper t o focus on the semantic component o f a n a t u r a l language system. For t h i s reason, the implementation has been r e s t r i c t e d as to s i z e and involvement with other aspects of n a t u r a l language a n a l y s i s . There are s e v e r a l dimensions along which the implementation c o u l d have been expanded. They are 1) To embed the semantic component i n a f u l l grammatical system, such as an ATN par s e r , and thus permit a more n a t u r a l i n p u t form. D i s c u s s i o n An,Implemented Model 69 2) To i n v e s t i g a t e the p o s s i b i l i t i e s f o r the use of r e s o l v e d prototypes and t r a c i n g i n f o r m a t i o n i n some subseguent l a n g u a g e - r e l a t e d behavior. .3) To expand the content of the pro t o t y p e , and s p e c i f y a s e t of case r o l e s to be f i l l e d and d e a l t with. H) To analyze the v a r i o u s c o n t r o l problems which a r i s e i n se a r c h i n g f o r r e s o l u t i o n s and t r a c i n g paths. The f i r s t three p o s s i b i l i t i e s were r e j e c t e d on the grounds t h a t they would r e q u i r e c o n s i d e r a b l e e f f o r t , only a s m a l l p a r t o f which would r e l a t e d i r e c t l y t o the i s s u e s beinq considered i n t h i s paper. Throughout the pre v i o u s s e c t i o n s , some i n d i c a t i o n s have been given as to how the s e a r c h i n g of prototypes could be c o n t r o l l e d . They a r e : 1) For any l e x i c a l element, order the prototypes by l e v e l of g e n e r a l i z a t i o n . 2) Allow f o r p o s s i b l e i n s t a n t i a t i o n of g e n e r a l i z e d prototypes which have a l r e a d y been r e s o l v e d . 3) Provide a measure by which t r a c e s or r e s o l u t i o n s may be compared. Avoid computation f o r r e s u l t s which c o u l d not be b e t t e r than "the best so f a r " . D i s c u s s i o n An Implemented Model 70 4) Do not continue to search f o r a r e s o l u t i o n (except by i n s t a n t i a t i o n ) i f a c e r t a i n l e v e l of a c c e p t a b i l i t y (low r e s o l u t i o n measure) has been met. To implement these c o n t r o l f e a t u r e s e f f e c t i v e l y , a l l components o f a t a r g e t prototype would have to be tr a c e d s i m i l t a n e o u s l y , and the sum of t h e i r t r a c e s would have to be c o n s t a n t l y monitored f o r t h r e s h h o l d s . In a d d i t i o n , a b r e a d t h - f i r s t search o f a l l the prototypes at the same l e v e l of g e n e r a l i z a t i o n would be d e s i r a b l e . T h i s would r e q u i r e t h a t the procedures which were attempting t r a c e s suspend themselves o c c a s i o n a l l y so t h a t c o n t r o l c o u l d be passed to other c o n s i d e r a t i o n s , or as an a l t e r n a t i v e , a l l such prototypes could be analysed i n p a r a l l e l . A l l i n a l l , i t was f e l t t h a t the implementation of the c o n t r o l s t r a t e g i e s would have been l a r g e l y a problem of the adequacy of programming language c o n s t r u c t i o n s , and again not c e n t r a l t o the theme of t h i s paper. To conclude t h i s s e c t i o n , mention should be made of some a r b i t r a r y d e c i s i o n s t h a t were made i n implementing the model. Depth of r e c u r s i o n i n t r a c i n g was not permitted beyond two l e v e l s (except f o r the use of ISA r e l a t i o n s ) . T h i s was done i n order that the examples of t r a c i n g would not i n c l u d e anything too obscure (which would have a p r o p o r t i o n a t l y high measure). The methods f o r measuring a r e s o l u t i o n were a l s o a r b i t r a r i l y imposed. The c r i t e r i a f o r the a c c e p t a b i l i t y of a t r a c e should D i s c u s s i o n An Implemented Model 71 most l i k e l y not be a const a n t , but r a t h e r be dependent on the context of the language use. One would expect a d i f f e r e n t degree of " s t r a i g h t - f o r w a r d n e s s " i f a s t r a n g e r stepped up to t a l k to you on the s t r e e t than you would reading someone's master's t h e s i s . D i s c u s s i o n Summary and Conclus i o n s 72 1. Summary, and Conclus i o n s T h i s paper has been motivated by the b e l i e f i n two p r i n c i p l e s concerning the a n a l y s i s of n a t u r a l language, 1) That a task-independent, bottom-up semantic a n a l y s i s i s p o s s i b l e to the point of the i d e n t i f i c a t i o n of knowledge r e l e v a n t t o an input, 2) Rather than d e v i s i n g methods f o r the r e p r e s e n t a t i o n of sentence meaning, emphasis should be placed on the o r g a n i z a t i o n and a p p l i c a t i o n of semantic knowledge. A semantic component of a n a t u r a l language a n a l y s i s system has been proposed which bears resemblances t o both case-based systems and t o semantic networks. The methods f o r determining the case f i l l i n g a b i l i t i e s of nouns has been extended beyond semantic markers to i n c l u d e a network o r g a n i z a t i o n of semantic knowledge. The case-frame i n f o r m a t i o n can be used to determine the f u l f i l l m e n t of other case-frames, and thus an i n c r e a s e d semantic p r o d u c t i v i t y has been provided. The concept of a semantic network has a l s o been extended t o permit the r e t e n t i o n o f semantic knowledge at s e v e r a l d i f f e r e n t l e v e l s . One weakness of semantic network r e p r e s e n t a t i o n s i s the Summary and C o n c l u s i o n s Summary and C o n c l u s i o n s 73 l a c k o f c o n t r o l on t h e p r o c e s s e s which s e a r c h t h e network.. T h i s c o n t r o l h as been p r o v i d e d by d i r e c t i n g t h e s e a r c h t o w a r d s t h e r e s o l u t i o n o f c a s s - f r a m e - l i k e p r o t o t y p e s . On the o t h e r hand, one weakness o f c a s e s y s t e m s - t h a t m u l t i p l e r e p r e s e n t a t i o n s a r e o f t e n n e c e s s a r y t o r e p r e s e n t s i m i l a r word u s e s - has been o vercome by t h e i n c l u s i o n o f a s e m a n t i c network i n t h e c a s e f i l l i n g c o n s i d e r a t i o n s . R u l e s f o r t h e use o f s e m a n t i c knowledge have been p r o p o s e d w h i c h c o v e r t h e u s u a l c a s e - f i l l i n g r e q u i r e m e n t s . I f , however, t h e r u l e s a r e r e l a x e d , and t h e s y s t e m i s c o m m i t t e d t o make t h e b e s t s e n s e p o s s i b l e o f any i n p u t , r a t h e r t h a n f i l t e r o u t a n o m a l i e s , i n t e r e s t i n g and c l e a r l y m e t a p h o r i c l a n g u a g e u s e s a r e a c c e p t e d , and t h e b e g i n n i n g s of an i n t e r p r e t a t i o n c an be d e v e l o p e d i n t e r m s o f t h e t r a c i n g o f t h e i n p u t components. One s i g n i f i c a n t a s p e c t o f t h e a p p r o a c h which has been d e v e l o p e d i i f t h i s p a p e r i s t h a t t h e c o n s i s t e n c y o f i n f o r m a t i o n i n r e l a t i o n a l n e t w o r k s , c a s e - f r a m e s , and p r i m i t i v e s p e c i f i c a t i o n has been i n v e s t i g a t e d . Some i n d i c a t i o n s have been p r o v i d e d as t o how t h e s e t o o l s o f n a t u r a l l a n g u a g e a n a l y s i s c o u l d be d e v e l o p e d i n harmony. I n a d d i t i o n , a d e g r e e o f c o m p a t i b i l i t y w i t h i n t u i t i o n a b o u t l a n g u a g e use has been m a i n t a i n e d . The p o s s i b i l i t y o f e x t e n d i n g t h e s y s t e m t o i n c o r p o r a t e t h e a b i l i t y t o l e a r n t h e s e m a n t i c s o f a l a n g u a g e t h r o u g h e x p e r i e n c e has been c o n s i d e r e d . Summary and C o n c l u s i o n s Program L i s t i n g s 74 8« Program L i s t i n g s Included here i s a l i s t i n g o f the r o u t i n e s c e n t r a l to the development of the examples shown above. Not i n c l u d e d are the data-base forming, and d i s p l a y r o u t i n e s . (DEFUN RESOLVE (INPUT) ( (LAMBDA (SUBJECT VERB OBJECT) ((LAMBDA (S-LIST V-LIST O-LIST) (RES0LVE1 V-LIST)) (CHAIN SUBJECT) (REVERSE (CONS VERB (CHAIN VERB))) (CHAIN OBJECT))) (SUBJECT INPUT) (VERB INPUT) (OBJECT INPUT))) (DEFUN RES0LVE1 (LIST) (COND ((NULL LIST) NIL) (T (APPEND (MAPCAR 'RESOLVE2 (PROTOTYPES (CAR LIST))) (RES0LVE1 (CDR L I S T ) ) ) ) ) ) (DEFUN RESOLVE2 (PROTO) ( (LAMBDA (S) (AND S ( (LAMBDA (0) (AND 0 ((LAMBDA (V) (LIST PROTO (LIST S V 0)) ) (COND ((EQ (VERB PROTO) VERB) «(IDENTITY)) (T (LIST 'ISA VERB Program L i s t i n g s Program L i s t i n g s 75 (VERB PROTO))))) (TRACK OBJECT (OBJECT PROTO) O-LIST 0 V-LIST)))) (TRACK SUBJECT (SUBJECT PROTO) S-LIST 0 V-LIST))) (DEFUN TRACK (SOURCE TARGET UP-LIST DEPTH ILLEGAL) (COND ( (GREATERP DEPTH 2) NIL) ((EQ TARGET SOURCE) ' (IDENTITY)) ( (MEMQ TARGET UP-LIST) (LIST 'ISA SOURCE TARGET)) ( (MEMQ SOURCE (CHAIN TARGET)) (LIST 'GEN SOURCE TARGET)) ((SHARE?)) ((SIMILAR?)) ((BOTHIN?)))) (DEFUN SHARE? NIL ((LAMBDA (X) (AND X (LIST ' SHARE SOURCE TARGET X) ) ) (INTERSECTQ (CHAIN SOURCE) (GET TARGET 'UP)))) (DEFUN SIMILAR? NIL ( (LAMBDA (X) (AND X (LIST 'SIMILAR SOURCE TARGET X))) ( (LAMBDA (Z) (MAPCAN • (LAMBDA (Y) (AND (EQ (SUBJECT Y) TARGET) (ASSOC (CONS SOURCE (CDAR Y) ) Z) (LIST (CDAR X) ) ) ) (PROTOTYPES TARGET) ) ) (PROTOTYPES SOURCE)))) (DEFUN BOTHIN? NIL (BOTHIN1? (CONS TARGET (CHAIN TARGET)))) (DEFUN BOTHIN1?) Program L i s t i n g s Program L i s t i n g s 76 (WORDS) (COND ( (NULL WORDS) NIL) ((EOTHIN2? (PROTOTYPES (CAR WORDS)))) (T (BOTHIN1? (CDR WORDS))))) (DEFUN BOTHIN2?) (PROTOS) (COND ((NULL PROTOS) NIL) ( (BOTHIN3? (CAR PROTOS)) ) ((BOTHIN 2? (CDR PROTOS))))) (DEFUN BOTHIN3?) (X) (AND (NULL (MEMQ (VERB X) ILLEGAL)) ( (LAMBDA (TRAC) (AND TRAC (APPEND (LIST 'APPEARS SOURCE TARGET X) (COND ( (MEMQ TARGET (CAR X)) (LIST TRAC)) (T (LIST (LIST 'ISA TARGET (OTHER 1 T R A C ) ) ) ) ) ) (TRACK SOURCE (OTHER TARGET X) UP-LIST (ADD1 DEPTH) (APPEND (CHAIN (VERB X)) (CONS (VERB X) I L L E G A L ) ) ) ) ) ) (DEFUN VERB) (INPUT) (CADAP INPUT)) (DEFUN SUBJECT) (INPUT) (CAAR INPUT) ) (DEFUN OBJECT) (INPUT) (CADD.AR INPUT) ) (DEFUN PROTOTYPES) (WORD) (GET WORD 'PROTOTYPES)) Program L i s t i n g s Program L i s t i n g s 77 (DEFUN CHAIN) (WORD) ({LAMBDA (X) (APPEND X (MA PC AN 'CHAIN X))) (GET WORD 'UP))) (DEFUN STRING) (WORD) (MAPCAN » (LAMBDA (X) (MAPCAR '{LAMBDA (Y) (CONS X Y) ) (OR (STRING X) • (NIL)))) (GET WORD 'UP))) (DEFUN OTHER) (TARGET PROTOTYPE) (COND ((MEMQ (SUBJECT PROTOTYPE) WORDS) (OBJECT PROTOTYPE) ) (T (SUBJECT PROTOTYPE)))) (DEFUN OTHER 1) (WORD PROT) (COND ( (EQ (SUBJECT PROT) WORD) (OBJECT PROT)) (T (SUBJECT PROT) ) )) (DEFUN EVALRES) (RES) .(APPLY 'ADD (REMOVE NIL (MAPCAR 'EVALRES1 (CADR RES))))) (DEFUN EVALRES1) (TRACE) (SELECTQ (CAR TRACE) (IDENTITY 0) (ISA (DISTANCE (CADR TRACE) (CADDR TRACE))) (GEN (ADD 3 (DISTANCE (CADDR TRACE) (CADR TRACE) ) ) ) (SHARE (MAX 6 (SUB 10 (TIMES (LENGTH (CADDDR TRACE)) 2)) )) (SIMILAR (MAX 8 (SUB 12 (TIMES (LENGTH (CADDDR TRACE)) 2)) )) (APPEARS (ADD 10 (APPLY 'ADD (MAPCAR »EVALRES1 Program L i s t i n g s Program L i s t i n g s 78 (CDDDDR TRACE))))) NIL)) (DEFUN DISTANCE) (S T) (APPLY 'MIN (MAPCAN '(LAMBDA (L) ( (LAMBDA (X) (AND X (LIST (ADD 1 (SUB (LENGTH L) (LENGTH X))) (MEMQ T L) )) (STRING S) ) ) ) Program L i s t i n g s B i b l i o g r a p h y 79 9* B i b l i o g r a p h y Anderson, J . and Bower, G. 1973. Human A s s o c i a t i v e Memory.. Winston-Wiley, Washington D.C. 1973 Bach, Emmon and Harms, Robert (eds.) 1968. U n i v e r s a l s i n L i n g u i s t i c Theory. H o l t , Rinehart and Winston, New York "1968 Becker, Joseph D. 1969. The Modeling of Simple A n a l o g i c and I n d u c t i v e Processes i n a Semantic Memory System. S t a n f o r d Al Memo No. 77, Jan 1969 Bobrow, Da n i e l G. and Winograd, Terry 1976. An Overview of KRL, a Knowledge Repres e n t a t i o n Language, mimeo 1976 Bobrow, D.G. 1975. Dimensions of Representation i n Bobrow and C o l l i n s , 1975 Bobrow, D a n i e l G. and C o l l i n s , A l l e n (eds.) 1975. Repre s e n t a t i o n and Understanding. New York, Academic Press, T975. "~ ~ B o l i n g e r , Dwight 1965. The Atomization of Meaning. Language v o l . 41, no.4 1965 pp. 555-573 Bruce, Bertram 1973. Case S t r u c t u r e Systems. IJCAI3 Stanford 1973 pp. 364-371. Bruce, Bertram 1975. Case Systems f o r N a t u r a l Language. A r t i f i c i a l ISiSliiafJice 6, 1975 pp. 327-36 0. C a r r o l l , John B. and F r e e d l e , Roy 0. (eds,) 1972. Language • comprehension and the A c g u i s i t i p n • of Knowledge. Washington, D.C., Winston~and Sons7 1972. C a r r o l l , John B. 1972. D e f i n i n g Language Comprehension: Some S p e c u l a t i o n s , i n C a r r o l l and F r e e d l e , 1972. Charniak, Eugene 1975. O r g a n i z a t i o n and Inference i n a Frame-Like System of . Common Sense Knowledge. i n Schank and Nash-Webber, 1975. pp.46-55. B i b l i o g r a p h y B i b l i o g r a p h y 80 Chomsky, Noam 1965. Aspects of the Theory of Syntax M.I.T., Cambridge, 1965 C o f e r , C h a r l e s N. (ed.) 1976. The S t r u c t u r e of Human .Memory.. San F r a n c i s c o , Freeman 1976. C o l l i n s , Alan M. and Q u i l l i a n , M. Boss 1972. How to Make a Language User. i n T u l v i n g and Donaldson, 1972. Davidson, James D. T o p i c s i n Discourse A n a l y s i s . . MSc T h e s i s , September 1976, U.B.C. F i l l m o r e , C h a r l e s J. 1976, The Case f o r Case Reopened. mimeo 1976, F i l l m o r e C h a r l e s , J . 19 74. The Future of Semantics. Berkley; S t u d i e s i n Syntax and Semantics. 1974 ~* ~ - - -F i l l m o r e , C h arles J. 1968. The Case f o r Case. i n Bach and Harms. 1968. Fodor, J.A. and Katz, J . J . (eds.) 1964. ghe S t r u c t u r e of Language. P r e n t i c e - H a l l , New J e r s e y 1964 G o l d s t e i n , I r a and Papert, Seymour 1976. A r t i f i c i a l I n t e l l i g e n c e , Language and the Study of Knowledge. MIT A l Lab memo 337, March 1976, Katz, J . J . and Fodor, J.A. 1963. The S t r u c t u r e of a Semantic Theory. i n Fodor and Katz 1964 pp. 479-518 McCawley, James D. 1968. The Role of Semantics i n a Grammar. i n Bach and Harms 1968 pp. 124-169 McDermott, Drew 1976. A r t i f i c i a l I n t e l l i g e n c e Meets N a t u r a l S t u p i d i t y . SIGART Newsletter No. 57 A p r i l 1976 pp. 4-9 Minsky, M. (ed.) 1968. Semantic Information P r o c e s s i n g . M.I.T. Press, Cambridge, 1968 Nash-Webber, Bonnie L. 1976. Semantic I n t e r p r e t a t i o n R e v i s i t e d . BBN Report 3335, J u l y 1976. B i b l i o g r a p h y B i b l i o g r a p h y 81 Norman, D.A. and Rumelhart, D.E. and the LNR Research Group 1975. E x p l o r a t i o n s i n C o g n i t i o n . Freeman and Co., San F r a n c i s c o 1975 Q u i l l i a n , M.R. 1968. Semantic Memory. i n Minsky, 1968 Raphael, Bertram 1968. SIR: A Computer Program f o r Semantic Information R e t r i e v a l . , i n Minsky, 1968 Rosch, Eleanor and Mervis, C a r o l y n B. et a l . 1976. B a s i c O b j e c t s i n Natu r a l C a t e g o r i e s . C o g n i t i v e Psychology. V o l . 8, No. 3 1976 pp. 382-439. Rumelhart, D.E. and Norman, D.A. 1973. A c t i v e Semantic Networks as a Model of Human Memory. IlSjyil S t a n f o r d 1973 pp. 450-457 Rumelhart, D.E. and L i n d s a y , P. H. and Norman, D.A. 1972. A Process Modal f o r Long-Term Memory, i n T u l v i n g and Donaldson, 1972 Schank,Roger C. 1976. The Role o f Memory i n Language P r o c e s s i n g . i n C o f e r , 1976. pp. 162-89. Schank,Roger C, 1975a, The S t r u c t u r e o f Episodes i n Memory. i n Bobrow and C o l l i n s , 1975. Schank, Roger C. 1975b. Using Knowledge to Understand. i n Schank and Nash-Webber, 1975. Schank, Roger C, 1972. Conceptual Dependancy: A Theory of N a t u r a l Language Understanding, C o g n i t i v e Psychology NO.3, 1972 pp.552-631 Schank,Roger C.and Colby, Kenneth M, (eds.) 1973. Computer Models of Thought and Language. San f r a n c i s c o , Freeman. 1976. Schank, Roger C. and Nash-Webber, Bonnie L, (eds.) 1975. T h e o r e t i c a l Issues i n N a t u r a l Language P r o c e s s i n g . Cambridge, Mass. 1975. Schubert, L.K. 1976. Extending the E x p r e s s i v e Power of Semantic Networks. A r t i f i c i a l I n t e l l i g e n c e Vol. 7 No. 2 1976, pp.163-198 B i b l i o g r a p h y B i b l i o g r a p h y 82 T a y l o r , Brock H. and Rosenberg, Richard S. 1975. A Case-Driven P a r s e r f o r Nat u r a l Language. AJCL M i c r o f i c h e 31, 1975. T u l v i n g , Endel and Donaldson, Wayne (eds.) 1972. 0£31Hization of Mejory;. N e w York, Academic P r e s s 7 1 9 7 2 7 W i l k s , Y o r i c k 1976. Frames, S c r i p t s , S t o r i e s and F a n t a s i e s . Proc. COLING 1976. W i l k s , Y o r i c k 1975a. An I n t e l l i g e n t Analyzer and Understander of E n g l i s h . Comm. ACM Vol.18, No. 5, 1975. pp. 264-274. Wi l k s , Yorick 1975b. A. P r e f e r e n t i a l , P a t t e r n - S e e k i n g , Semantics f o r Natu r a l Language Inference, A r t i f i c i a l I n t e l l i g e n c e 6, 1975. pp. 53-74. W i l k s , Y o r i c k 1975c. P r i m i t i v e s and Words i n Schank and Nash-Webber, 1975. Wilks, Y o r i c k 1975d. Seven Theses on A l and N a t u r a l Language. I s t u t o per g l i S t u d i Semantici e C o g n i t i o , No. 17, 1975 Wil k s , Y o r i c k 1975e. Methodology i n A l and Nat u r a l Language Understanding, i n Schank and Nash-Webber, 1975: a l s o Wilks, 1975d Wi l k s , Yorick 1974., Natu r a l Language Understanding Systems Within the A l Paridigm. S t a n f o r d A l Lab memo 237 December 1974 Wi l k s , Y o r i c k 1973a. An A r t i f i c i a l I n t e l l i g e n c e Approach to Machine T r a n s l a t i o n . i n Schank and Colby, 1973. Wil k s , Y o r i c k 1973b. Preference Semantics. Stanford Al Lab Memo 206, J u l y 1973 Wilks, Y o r i c k 1973c. N a t u r a l Language Inference. Stanford A l Lab Memo 241, August 1973 Winograd, T e r r y 19 72. Understanding Natural Language. New York, Academic Press, 1972. ~ Bi b l i o g r a p h y B i b l i o g r a p h y 83 Winograd, T e r r y 1971, Procedures as a Representation f o r Data i n a Computer Program f o r Understanding N a t u r a l Language, M.I.T. AI-TR-17, February, 1971 Woods, W.A. 1968, P r o c e d u r a l Semantics f o r a Question-Answering Machine. Proceedings AF1PJ3 Conference V o l . 33 pp.457-471 B i b l i o g r a p h y 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            data-media="{[{embed.selectedMedia}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.831.1-0051787/manifest

Comment

Related Items