Semantic Networks and the Generation of Context by John ...

Semantic N e t w o r k s and t h e G e n e r a t i o n o f C o n t e x t by John M y l o p o u l o s ,

P h i l i p Cohen, A l e x a n d e r B o r g i d a and L a s z l o Sugar Department o f Computer S c i e n c e U n i v e r s i t y of Toronto T o r o n t o , C anada 416-928-5180

and g r a d e s , a p p l i c a t i o n s t o u n i v e r s i t i e s and decisions by u n i v e r s i t i e s regarding admission, etc. We c o n s i d e r t h e e d u c a t i o n a l w o r l d an example o f a n i n s t i t u t i o n a l w o r l d and r e g a r d f u r t h e r work on modelling p a r t i c u l a r i n s t i t u t i o n a l worlds q u i t e i m p o r t a n t s i n c e t h e y subsume many s i t u a t i o n s w h i c h arise in d a i l y l i f e , e.g. business. With the e x c e p t i o n o f work b y M a r t i n [ 5 ] and M a l h o t r a [ 6 , w e a r e n o t aware o f o t h e r m a j o r e f f o r t s t o r e p r e s e n t and use knowledge t h a t can be c l a s s i f i e d as institutional.

Abstract T h i s paper o u t l i n e s a r e p r e s e n t a t i o n of knowledge based o n s e m a n t i c n e t w o r k s and o r g a n i z e d i n terms o f s e m a n t i c axes and " s c e n a r i o s " . A cont e x t mechanism f o r such n e t s i s a l s o d i s c u s s e d , a l o n g w i t h a n a l g o r i t h m f o r i n t e g r a t i n g new i n f o r mation to a semantic n e t . F i n a l l y , a n example i s g i v e n t o i l l u s t r a t e t h e r e p r e s e n t a t i o n and t h e c o n t e x t mechanism. 0.

Introduction We must emphasize f r o m t h e b e g i n n i n g t h a t t h i s w o r k has a s g o a l t h e c r e a t i o n o f " i n t e l l i g e n t " computer systems r a t h e r t h a n c o g n i t i v e s i m u l a t i o n o f humans. The work was c a r r i e d o u t a s p a r t o f t h e TORUS (TORonto U n d e r s t a n d i n g System) p r o j e c t whose aim i s t o p r o v i d e n a t u r a l language f r o n t ends t o ( r e l a t i o n a l ) d a t a base management s y s t e m s . A p r o t o t y p e v e r s i o n of TORUS has been i m p l e m e n t e d , ( M y l o p o u l o s e t a l [ 7 ] , and uses a s i m p l e r v e r s i o n o f t h e s e m a n t i c memory d e s c r i b e d h e r e . The universe o f discourse f o r the p r o t o t y p e i s , a g a i n , the educational w o r l d .

The r e p r e s e n t a t i o n o f knowledge i s one o f t h e most i m p o r t a n t p r o b l e m s o f A I t o d a y . Early approaches t o A I a t t e m p t e d t o s i m p l i f y t h e p r o b l e m s tackled by proposing syntactic solutions to i n h e r e n t l y semantic problems, thus u n d e r s t a t i n g the importance of the r e p r e s e n t a t i o n of knowledge. The h i s t o r y o f A I has t a u g h t u s , h o w e v e r , t h a t t h e r e are no s h o r t c u t s to the process of " u n d e r s t a n d i n g " . T h i s p r o c e s s can o n l y be a c c o m p l i s h e d by a s y s t e m t h a t possesses knowledge about t h e t o p i c i t i s expected to understand. M o r e o v e r , t h e degree o f u n d e r s t a n d i n g i s p r o p o r t i o n a l t o t h e degree t o w h i c h t h i s knowledge can b e used f o r r e c o g n i t i o n and p r e d i c t i o n . The i m p o r t a n c e o f t h e r e p r e s e n t a t i o n o f knowledge stems f r o m t h e s e p r e m i s e s .

1.

World M o d e l l i n g

By " w o r l d m o d e l l i n g " we mean h e r e t h e p r o c e s s o f r e p r e s e n t i n g t h e knowledge r e l e v a n t t o a t o p i c of discourse. In t h i s section we introduce the TORUS r e p r e s e n t a t i o n and g i v e s e v e r a l examples o f how knowledge a b o u t a domain of d i s c o u r s e can be represented.

I n t h i s p a p e r w e o u t l i n e a model o f s e m a n t i c r e p r e s e n t a t i o n t h a t i n c l u d e s a semantic net(work) and a c o n t e x t mechanism, and i l l u s t r a t e i t s use w i t h s e v e r a l examples. Our s e m a n t i c n e t i s o r g a n i z e d a r o u n d t h e concept of a " s c e n a r i o " , i . e . a semantic u n i t t h a t may encompass more t h a n j u s t i s o l a t e d p r o p o s i t i o n s . T h i s i s i n agreement w i t h c u r r e n t t r e n d s i n t h i s a r e a o f r e s e a r c h (Minsky [ 1 ] , Schank [ 2 ] , A b e l s o n [3] etc.)* S c e n a r i o s a s w e l l a s more e l e m e n t a r y semantic e n t i t i e s are organized on the semantic net a l o n g two axes we have c a l l e d t h e SUB and DEF a x e s . The SUB a x i s has been used e l s e w h e r e (Norman and R u m e l h a r t [ 4 ] , M a r t i n I 5 ] ) and a f o r m o f t h e DEF a x i s has a l s o been used i n [ 4 l .

1.1

The Semantic Network - B a s i c B u i l d i n g B l o c k s

The s e m a n t i c n e t w o r k i s s i m p l y a l a b e l l e d d i r e c t e d g r a p h , where b o t h edges and nodes may be l a b e l l e d . T h e l a b e l s o f nodes w i l l o n l y b e used f o r r e f e r e n c e purposes and w i l l u s u a l l y b e mnemonic names. The l a b e l s o f e d g e s , o n t h e o t h e r h a n d , w i l l have a number o f a s s o c i a t e d s e m a n t i c p r o p e r t i e s and i n f e r e n c e s . I t must b e n o t e d t h a t o u r use o f t h e t e r m "semantic network" i s d i f f e r e n t from t h a t o f Q u i l l i a n [ 8 ] , Simmons [ 9 ] , e t c . However, w e f e e l t h a t t h e t e r m c a p t u r e s t h e essence o f t h e o b j e c t i t names and p r o p o s e t o c o n t i n u e t o use i t .

Given a semantic n e t , we c o n s i d e r t h e problem o f d e f i n i n g and u s i n g a c o n t e x t mechanism f o r r e f e r e n t d e t e r m i n a t i o n , (semantic) d i s a m b i g u a t i o n and i n f e r e n c e . The p r o b l e m o f i n t e g r a t i n g new i n f o r m a t i o n t o t h e semantic net i s a l s o t r e a t e d t h r o u g h a p r o c e s s w e c a l l " g r a p h f i t t i n g " and t h e r e l a t i o n s h i p o f c o n t e x t and g r a p h f i t t i n g i s d i s cussed. F i n a l l y , a l o n g example i s p r e s e n t e d t o i l l u s t r a t e both the s u i t a b i l i t y o f the representat i o n and t h e use o f t h e c o n t e x t mechanism.

We r e c o g n i z e t h r e e b a s i c t y p e s of n o d e s : c o n c e p t s , e v e n t s and c h a r a c t e r i s t i c s , w h i c h a r e used t o r e p r e s e n t t h e i d e a s making u p o u r u n i v e r s e . Concepts are t h e e s s e n t i a l c o n s t a n t s o r p a r a m e t e r s o f o u r w o r l d and s p e c i f y o b j e c t s o r abstractions. There i s a l o o s e c o n n e c t i o n b e t w e e n c o n c e p t s and a s u b s e t o f t h e c l a s s o f n o u n s .

The knowledge we have r e p r e s e n t e d and used f o r most o f t h e examples i s d e r i v e d f r o m t h e " e d u c a t i o n a l w o r l d " , i . e . the w o r l d o f s t u d e n t s , courses

134

E v e n t s a r e used t o r e p r e s e n t t h e a c t i o n s w h i c h can o c c u r i n t h e w o r l d t h a t w e a r e m o d e l l i n g . Their representation ased on a case-grammar m o d e l , w i t h w h i c h w e assume t h e r e a d e r i s f a m i l i a r , cases (or r o l e s ) w i l l b e r e p r e s e n t e d b y c a s e l a b e l l e d edges r a d i a t i n g o u t w a r d f r o m t h e e v e n t node and p o i n t i n g t o t h e nodes w h i c h f i l l t h e r e s p e c t i v e cases.

c h a r a c t e r i s t i c o f a PERSON, a l t h o u g h , i f c o n s i d e r e d i n more d e t a i l , i t c o u l d b e r e p r e s e n t e d a s t h e LOCATION of t h e DWELLING w h i c h t h e PERSON u s u a l l y INHABITS. I n some c i r c u m s t a n c e s such a b b r e v i a t i o n s a r e mappings f r o m a c r o s s - p r o d u c t domain to a r a n g e and we use a WRT ( w i t h - r e s p e c t - t o ) edge to i n d i c a t e t h e second a r g u m e n t . For e x a m p l e , GRADE c h a r a c t e r i z e s STUDENTS w i t h r e s p e c t to COURSE, p r o d u c i n g a GRADE.VALUE:

For e x a m p l e ,

represents an instance of the event "apply" w i t h t h e a g e n t case f i l l e d b y " j o h n " and t h e d e s t i n a t i o n case f i l l e d b y " d e p a r t m e n t " . The f o l l o w i n g i s t h e l i s t o f t h e cases presently in use: agent ( a ) , a f f e c t e d ( a f f ) , t o p i c ( t ) , i n s t r u m e n t ( i ) , r e s u l t ( r ) , source ( s ) , d e s t i n a t i o n (d) and o b j e c t ( o ) . T h e i r names a r e intended to be approximately self-explanatory.

In a d d i t i o n to these types of e n t i t i e s , we w i l l sometimes use m a t h e m a t i c a l p r e d i c a t e s such a s MEMBER, DIFFERENCE, e t c .

For a complete u n d e r s t a n d i n g of an e v e n t , a l l o f i t s cases must b e known. We w i l l not i n s i s t , h o w e v e r , t h i s a l w a y s b e done a t t h e t i m e t h e e v e n t i s f i r s t presented f o r i n t e g r a t i o n w i t h the semantic n e t and t h e c o n t e x t . Many s e m a n t i c cases w h i c h a r e o m i t t e d f r o m t h e i n p u t a r e f i l l e d i n b y d e f a u l t and c o n t e x t u a l l y i n f e r r e d cases when t h i s i n p u t i s i n t e g r a t e d i n t o the net. The cases o f a n e v e n t can o f t e n b e r e l a t e d t o l i n g u i s t i c s y n t a c t i c phenomena ( e . g . p r e p o s i t i o n s , i n f l e c t i o n a l a f f i x e s , e t c . ) of a corresponding verb and t h i s c o r r e l a t i o n can b e v e r y h e l p f u l i n o b t a i n i n g t h e mapping between a c t u a l s e n t e n c e s and t h e i r semantic r e p r e s e n t a t i o n .

The nodes t h a t c o n s t i t u t e t h e s e m a n t i c n e t w i l l b e d i v i d e d i n t o two c l a s s e s : one, r e l a t i n g t o g e n e r i c c o n c e p t s , e v e n t s and c h a r a c t e r i s t i c s , describes the p o s s i b l e or allowable s t a t e of a f f a i r s i n o u r domain o f d i s c o u r s e . This c l a s s we s h a l l i n f o r m a l l y c a l l t h e " u p s t a i r s " o f t h e semant i c n e t w o r k i n c o n t r a s t t o t h e second c l a s s , t's " d o w n s t a i r s " , where w e keep t h e i n s t a n t i a t i o n s and p a r t i c u l a r occurences o f v a r i o u s i d e a s . "Upstairs" nodes w i l l have t h e i r names g i v e n i n c a p i t a l l e t t e r s , whereas nodes " d o w n s t a i r s " w i l l b e i n lower-case.

A more i m p o r t a n t f u n c t i o n o f cases i s t o a i d i n i n f e r e n c e and o t h e r l o g i c a l o p e r a t i o n s w h i c h must b e p e r f o r m e d when i s o l a t e d e v e n t s a r e p u t i n t o c o n t e x t . T h i s i s a n i m p o r t a n t t o p i c and w i l l b e discussed in section 2.2. C h a r a c t e r i s t i c s a r e used t o r e p r e s e n t s t a t e s and t o m o d i f y c o n c e p t s , e v e n t s o r o t h e r c h a r a c t e r istics. F o r c o n c e p t s , c h a r a c t e r i s t i c s can b e q u a l i t i e s , t r a i t s o r p r o p e r t i e s which are i n h e r e n t i n , o r ascribed t o , the concept. For e v e n t s , c h a r a c t e r i s t i c s describe g e n e r a l p r o p e r t i e s which a r e l e s s c l o s e l y a s s o c i a t e d w i t h t h e e v e n t t h a n are cases and w h i c h have a more g e n e r a l , u n i f o r m meaning a t a l l t i m e s ( e . g . l o c a t i o n , t i m e , d u r a t i o n , etc.). A c h a r a c t e r i s t i c may be c o n s i d e r e d to be a b i n a r y r e l a t i o n , mapping e l e m e n t s f r o m i t s domain t h o s e nodes t o w h i c h t h e c h a r a c t e r i s t i c may a p p l y , to t h e range - those values which t h e c h a r a c t e r i s t i c may t a k e . F o r e x a m p l e , SEX maps ANIMAL i n t o t h e s e t c o n t a i n i n g MALE, FEMALE. G r a p h i c a l l y , the c h a r a c t e r i s t i c i s r e p r e s e n t e d a s a node l a b e l l e d b y t h e name o f t h e c h a r a c t e r i s t i c , w i t h a " c h " ( " c h a r a c t e r i z e " ) edge p o i n t i n g t o a n e l e m e n t o f t h e domain and a " v " ( " v a l u e " ) edge p o i n t i n g t o t h e corresponding value:

The a p p a r a t u s d e s c r i b e d i n t h e p r e v i o u s s e c t i o n i s s u f f i c i e n t f o r the r e p r e s e n t a t i o n o f most i s o l a t e d phenomena, b u t w e need t h e a b i l i t y t o c o n s t r u c t l a r g e r u n i t s o f knowledge w h i c h w i l l help in inference, referent determination, prediction, etc. Such s t r u c t u r e s have been p r o p o s e d i n v a r i o u s forms b y many o t h e r s , i n c l u d i n g M i n s k y ' s frames [ 1 ] , S c h a n k ' s s c r i p t s [ 2 ] , R i e s b e c k ' s s t o r y patterns [10], etc. W e c a l l o u r v e r s i o n o f such structures scenarios. A scenario is a c o l l e c t i o n of events, charact e r i s t i c s and m a t h e m a t i c a l p r e d i c a t e s r e l a t e d t h r o u g h t h e i r a s s o c i a t e d edges and t h e c o n c e p t s f i l l i n g t h e i r cases, o r through temporal and/or c a u s a l c o n n e c t i v e s , such a s " b e f o r e " , " a f t e r " , " s a m e . t i m e " , " w h i l e " , " p r e r e q u i s i t e " and " e f f e c t " . One can r e g a r d a s c e n a r i o as a p a t t e r n w h i c h , when matched b y a s t r u c t u r e , e n a b l e s t h e s y s t e m t o make i n f e r e n c e s and p r e d i c t i o n s . Consider, for example, the n o t i o n of "a p e r s o n ' s c u r r e n t a d d r e s s " , w h i c h may be r e p r e s e n t e d as

" T r u e " c h a r a c t e r i s t i c s are u s u a l l y n a t u r a l a t t r i b u t e s o f c o n c e p t s , b u t c h a r a c t e r i s t i c s can a l s o b e used a s a b b r e v i a t i o n s o f more c o m p l i c a t e d s i t u a t i o n s , where w e want t o o m i t u n n e c e s s a r y detail. Thus ADDRESS may be t h o u g h t of as a

135

"dimensions*' and s e v e r a l o t h e r types of edges. 1.3.1

The SUB a x i s

We w i l l say t h a t the node X is a SUBnode of node Y, if the s e t of i n s t a n t i a t i o n s of X is a subset of the s e t of i n s t a n t i a t i o n s of Y. The SUB r e l a t i o n between X and Y w i l l be denoted by " d o w n s t a i r s " , the r e l a t i o n between Y and of i n s t a n t i a t i o n or example-of. We w i l l to use an u n l a b e l l e d edge to denote such s i n c e the f a c t t h a t X is " d o w n s t a i r s " is s p e c i f i e d by i t s name. Thus

X is one continue relations, already

s p e c i f i e s t h a t STUDENT is a s u b c o n c e p t of PERSON, " J o h n . s m i t h " i s a n i n s t a n t i a t i o n o f PERSON ( i . e . a p e r s o n ) and " j i m . b r o w n " is a s t u d e n t . The SUB and i n s t a n t i a t i o n edges have been used by many o t h e r r e p r e s e n t a t i o n s of knowledge. We can now o r g a n i z e ( p a r t i a l l y o r d e r ) t h e c o n c e p t s o c c u r i n g i n o u r domain o f d i s c o u r s e i n t o a h i e r a r c h y , r e p r e s e n t a b l e b y i t s Hasse d i a g r a m . I t i s i m p o r t a n t t o note t h a t (semantic) p r o p e r t i e s o f c o n c e p t s a r e i n h e r i t e d a l o n g t h e SUB a x i s . T h u s , f o r e x a m p l e , s i n c e STUDENTS a r e PERSONS and PERSONS a r e PHYSICAL.OBJECTS, STUDENTS may have a WEIGHT a s s o c i a t e d w i t h t h e m . This property of the SUB a x i s p r o v i d e s us w i t h a v e r y i m p o r t a n t memorysaving device.

Here, " I t " stands f o r the numerical p r e d i c a t e " l e s s t h a n * , and " b v a r " s t a n d s f o r a bound v a r i a b l e whose v a l u e w e d o n o t know. C l e a r l y t h i s is a very s i m p l e s c e n a r i o s i n c e i t does n o t a l l o w any i n f e r e n c e s o r any ( n o n - t r i v i a l ) p r e d i c t i o n s . C o n s i d e r now a s c e n a r i o f o r recommending a person to an o r g a n i z a t i o n by w r i t i n g a l e t t e r :

S c e n a r i o s a r e a l s o o r g a n i z e d o n t h e SUB a x i s . C o n s i d e r t h e g e n e r i c c h a r a c t e r i s t i c f o r ADDRESS. LEGAL.PERSON

LEGAL.PERSON s t a n d s f o r a p e r s o n o r a n o r g a n i z a t i o n ( a n y t h i n g y o u can sue . . . ) . Below i t o n t h e SUB a x i s w e can p l a c e s c e n a r i o s f o r o f f i c e a d d r e s s a n d home a d d r e s s , f o r p e r s o n s o n l y :

T h i s s c e n a r i o d e s c r i b e s a recommendation a c t i o n i n t e r m s o f WRITE and TRANSFER a c t i o n s , t h e f i r s t being the p r e r e q u i s i t e o f the second. The r e l a t i o n s h i p between t h e cases o f t h e two a c t i o n s is also defined. Note t h a t t h e recommending p e r s o n is b o t h t h e a g e n t of WRITE and t h e AGENT-SOURCE OF TRANSFER. The names of t h e nodes h e r e i n d i c a t e t h e t y p e s o f i n s t a n t i a t i o n s t h a t can f i l l t h e cases o f this scenario. The " d i f f e r " edge s p e c i f i e s t h a t a p e r s o n c a n n o t recommend h i m s e l f . The r e a d e r may wonder how t h i s s c e n a r i o i s r e l a t e d t o RECOMMEND. We w i l l discuss o v e r a l l o r g a n i z a t i o n of the semantic net in the next sect i o n and w e o n l y w i s h t o m e n t i o n h e r e t h a t t h i s s c e n a r i o may b e t h o u g h t o f a s a n a b s t r a c t p r o c e d u r e d e f i n i n g RECOMMEND. T h e r e may be o t h e r s c e n a r i o s w h i c h d e f i n e WRITE and TRANSFER i n t e r m s o f o t h e r , more p r i m i t i v e , e v e n t s . 1.3

ADDRESS.VALUE

Thus a PERSON'S o f f i c e ADDRESS, d e f i n e d by t h e node marked w i t h * , i s t h e l o c a t i o n v a l u e ( a d d r e s s values are a l s o l o c a t i o n values i n t h i s r e p r e s e n t a t i o n ) of t h e OFFICE USEd by t h e PERSON. The r e p r e s e n t a t i o n o f USE and OFFICE i s r a t h e r s u p e r f i c i a l , b u t l e g a l , and w i l l d o t o d r i v e t h e p o i n t across.

S e m a n t i c Net O r g a n i z a t i o n

W e c o u l d now f i t t h e c u r r e n t a d d r e s s b e l o w the o f f i c e address s c e n a r i o , i n order t o d e f i n e the n o t i o n o f " c u r r e n t o f f i c e address"

S o f a r w e have d i s c u s s e d t h e e n t i t i e s t h a t serve as b u i l d i n g blocks f o r our r e p r e s e n t a t i o n . I n t h i s s e c t i o n w e d e s c r i b e how t h e y a r e p u t t o g e t h e r to form the semantic n e t . This organizat i o n w i l l b e d e f i n e d i n terms o f " a x e s " o r

136

The l e f t s u p e r s c r i p t 1 ' s i n d i c a t e t h e c o r r e s pondence o f nodes between t h e l a s t two f i g u r e s . Note t h a t t h e ADDRESS node marked w i t h * does n o t need a " c h " o r " v " e d g e , s i n c e t h e s e a r e i n h e r i t e d f r o m i t s SUPernode.

There need n o t b e j u s t a s i n g l e " d e f " o r " f e d " edge l e a v i n g a n e v e n t b e i n g d e f i n e d . There can b e s e v e r a l d e f i n i n g s c e n a r i o s f o r t h e same e v e n t and many p o s s i b l e e n t r y p o i n t s ( " d e f " edges) and e x i t p o i n t s ( " f e d " e d g e s ) . Moreover, an event o r c h a r a c t e r i s t i c may b e d e f i n e d r e c u r s i v e l y .

B e f o r e w e p r o c e e d t o d i s c u s s o t h e r axes used t o organize the semantic n e t , i t i s w o r t h w h i l e t o m e n t i o n a few t h i n g s a b o u t t h e t o p of t h e SUB a x i s , Every node o n t h e s e m a n t i c n e t f i t s b e l o w t h e node

The RECOMMEND e v e n t of t h e p r e v i o u s d i a g r a m may b e i t s e l f p a r t o f a n A P P L Y - t o - a n - o r g a n i z a t i o n s c e n a r i o , w h i c h can have a s SUBscenario A P P L Y - t o - a -university. T h i s , i n t u r n , may b e p a r t o f a " g o i n g t o c o l l e g e " s c e n a r i o , w h i c h i n t u r n may b e p a r t of a " g e t t i n g an education" scenario. These s c e n a r i o s a r e t h e n r e l a t e d t o each o t h e r a l o n g t h e DEF and SUB a x i s as f o l l o w s :

OBJECT,

I t i s i n t e r e s t i n g t o note scenarios being present at the o n l y s t r u c t u r e s which f a i l t o r u l e s o f the semantic net w i l l w i l l match n o s c e n a r i o o n t h e structure

t h a t w i t h such t o p o f t h e SUB a x i s , s a t i s f y the s y n t a c t i c be rejected ( i . e . net). Thus t h e

w i l l match a t l e a s t a t t h e most g e n e r a l c h a r a c t e r i s t i c scenario given e a r l i e r , provided that "color" can be c l a s s i f i e d as a c h a r a c t e r i s t i c and 3kg as a characteristic value. This f e a t u r e of the semantic net enables u s t o i n t r o d u c e scenarios t h a t w i l l account f o r ( s e m a n t i c a l l y ) unusual s t r u c t u r e s . These s c e n a r i o s may cause i n f e r e n c e s such as "I d o n ' t understand what h e ' s t a l k i n g a b o u t " , e t c . , and r e s u l t i n c o r r e s p o n d i n g a c t i o n s o n t h e p a r t o f the system ( q u e s t i o n the u s e r , e t c . ) . 1.3.2

The DEF a x i s

We have a l r e a d y d e f i n e d RECOMMEND ( t h r o u g h a l e t t e r ) in terms of a s c e n a r i o . We w i l l denote t h e r e l a t i o n s h i p between RECOMMEND and i t s c a s e s , on one h a n d , and t h e RECOMMEND s c e n a r i o , on t h e o t h e r h a n d , b y u s i n g " d e f " and " f e d " - l a b e l l e d edges:

A t one end o f t h e DEF a x i s one w o u l d f i n d v e r y g e n e r a l s c e n a r i o s and p l a n s . In the "educat i o n " w o r l d , these might i n v o l v e " g e t t i n g an education", "choosing a v o c a t i o n " , e t c . a c t i v i t i e s w h i c h i n t e r a c t w i t h a huge number o f o t h e r general scenarios which would lead o f f i n t o diverse areas. A domain o f d i s c o u r s e , f o r w h i c h our r e p r e s e n t a t i o n might be adequate, would a t t e m p t t o m i n i m i z e i t s dependence upon t h i s e n d of the d e f i n i t i o n a l a x i s . One o f t h e u n d e r l y i n g a s s u m p t i o n s o f any a t t e m p t t o model s m a l l , c l o s e d domains o f d i s c o u r s e i s t h a t t h i s m i n i m i z a t i o n can be d o n e . A t t h e o t h e r end o f t h e DEF a x i s , one w o u l d f i n d " p r i m i t i v e s " — those e n t i t i e s which we choose n o t t o r e p r e s e n t and t h u s t a k e a s " g i v e n " , i m p l y i n g t h a t t h e y a r e , somehow, u n d e r s t o o d . U n l i k e Schank, w e b e l i e v e t h a t t h e p r i m i t i v e s t h a t would be i n c o r p o r a t e d i n t o a net should be depend e n t upon t h e domain o f d i s c o u r s e a s w e l l a s convenience. F i n a l l y , w e s h o u l d m e n t i o n t h a t a n o t h e r edge w h i c h d e f i n e s an a x i s is t h e " p a r t " edge (a GRADE i s a " p a r t " o f a TRANSCRIPT, w h i c h i s a " p a r t " o f an APPLICATION, a DEPARTMENT is a " p a r t " of a UNIVERSITY, e t c . ) . 1.4

P r o c e d u r a l Knowledge

R e p r e s e n t i n g knowledge o n t h e s e m a n t i c n e t has t h e advantage t h a t t h i s i n f o r m a t i o n can b e examined and reasoned a b o u t by t h e s y s t e m . On the other hand, the r e p r e s e n t a t i o n is expensive.

137

and f o r any u n i v e r s e o f d i s c o u r s e , t h e r e w i l l always b e ' p e r i p h e r a l ' knowledge f o r w h i c h g e n e r a l r e a s o n i n g may n o t be n e c e s s a r y We r e p r e s e n t such knowledge i n terms o f f u n c t i o n s w h i c h w e a s s o c i a t e t o c o r r e s p o n d i n g nodes o f t h e s e m a n t i c n e t . Some o f t h e s e f u n c t i o n s w e w i l l c a l l r e c o g n i t i o n f u n c t i o n s because t h e i r j o b i s t o use s y n t a c t i c and s e m a n t i c i n f o r m a t i o n i n o r d e r t o recognize instances of a class ( e . g . address.value). Mapping f u n c t i o n s w i l l b e u s e f u l f o r mapping s t r u c t u r e s f r o m one l e v e l o f t h e r e p r e s e n t a t i o n t o another. F o r e x a m p l e , mapping f u n c t i o n s may be used t o r e p l a c e p o r t i o n s o f t h e d e f i n i t i o n a l a x i s i f i t i s d e c i d e d t h a t o t h e r l e v e l s , more s u r f a c e o r deeper o n e s , a r e n o t o f i n t e r e s t . Another type of f u n c t i o n i s a d e f i n i t i o n a l f u n c t i o n / w h i c h i s used t o define procedures f o r performing p a r t i c u l a r a c t i o n s (EVALUATE an a p p l i c a t i o n , MOVE a b l o c k , etc.). I t i s i m p o r t a n t t o s t r e s s t h a t knowledge can be represented, in general, in e i t h e r procedural o r d e c l a r a t i v e f o r m and w h i c h f o r m i s s e l e c t e d i s s t r i c t l y an issue of t r a d i n g cost f o r 'understanding power'.

(partial)

scenario

and t h e s e n t e n c e s "John a p p l i e d y e s t e r d a y .

The g r a d e s were g o o d . "

We can g e n e r a t e p a r t of t h e c o n t e x t of APPLY b y i n t e r p r e t i n g t h e " r e s u l t " case and t h e " p a r t " edge a s d e s c r i b e d b e l o w . When APPLY o c c u r s , i t p r o d u c e s a n APPLICATION w h i c h e n t e r s t h e c o n t e x t . When w e b r i n g APPLICATION i n , w e b r i n g w i t h i t , i t s constituent "part"s. Thus GRADE and TRANSCRIPT are a v a i l a b l e f o r d i s c u s s i o n w i t h o u t a previous r e f e r e n c e , a s i s i n d i c a t e d b y the d e f i n i t e s p e c i f i c a t i o n i n t h e above e x a m p l e . Inferences are h e u r i s t i c a l l y assigned s t r e n g t h s which are used t o l i m i t t h e e x t e n t o f t h e i r p r o p a g a t i o n ( c f . R i e g e r [12 1) . Below we l i s t a few edge l a b e l s and d i s c u s s how c o n t e x t g r o w t h i s a f f e c t e d b y e a c h .

C o n t e x t and I t s Uses A is an e v e n t ( s c e n a r i o ) and B is a c o n c e p t . T h i s edge c o r r e s p o n d s t o t h e o u t p u t o f some process ( s c e n a r i o ) . T h u s , when A i s i n s t a n t i a t e d , an i n s t a n c e of B s h o u l d be c r e a t e d and added to the context. The p r o p a g a t i o n v a l u e o f t h i s e d g e , i n b o t h d i r e c t i o n s , i s h i g h , a s can b e seen f r o m t h e APPLY example a b o v e . Inferences from " r e s u l t " can o c c u r when B i s a c o n c e p t w h i c h i s e i t h e r a n " i n s t r u m e n t " o r " o b j e c t " used b y some o t h e r e v e n t C (more g e n e r a l l y , a n i n p u t t o C ) . This produces a p r e r e q u i s i t e o r d e r i n g of e v e n t s . Namely, A must be executed b e f o r e C since A ' s o u t p u t is i n p u t to C. T h i s i n f o r m a t i o n may be r e d u n d a n t l y d i s c o v e r e d from the " p r e r e q u i s i t e " edge, but the " r e s u l t i n p u t " i n f e r e n c e i s s t r o n g e r s i n c e i t e x p l a i n s why t h e p r o d u c e r i s p r e r e q u i s i t e t o t h e consumer.

I n t h i s s e c t i o n , w e d e s c r i b e how c o n t e x t can b e o r g a n i z e d s o t h a t immediate i n f e r e n c e s a r e c o n t e x t u a l l y a p p r o p r i a t e and v i c e v e r s a . The purpose o f the c u r r e n t c o n t e x t f a c i l i t y i s t o maintain information relevant to the topic at hand. T h i s i n f o r m a t i o n i s d e r i v e d from expectat i o n s and i n f e r e n c e s and uses s c e n a r i o s . The s e c t i o n below d e s c r i b e s t h e p r o c e s s o f g e n e r a t i n g a c o n t e x t , u s i n g i t f o r i n t e g r a t i n g i n p u t and determining sentential referents. We also discuss f i n d i n g the context r e l e v a n t to an i n p u t by a process c a l l e d "graph f i t t i n g " . Finally, a d e t a i l e d example i s p r e s e n t e d , s e r v i n g t o c o v e r many o f t h e s a l i e n t f e a t u r e s i n t h e p r e c e d i n g discussion. 2.1

The p r e s e n c e o f a n e t w o r k e n t i t y i n t h e c o n t e x t r e p r e s e n t s the system's e x p e c t a t i o n t h a t t h i s item is or w i l l be relevant to the current dialogue. When new i n f o r m a t i o n , w h i c h has been p r e d i c t e d , e n t e r s t h e d i a l o g u e , i t s r e l e v a n c e can be explained by the " g e n e r a t i o n p a t h " taken to create the e x p e c t a t i o n . We g i v e an example of p a r t o f the c o n t e x t g e n e r a t i o n process below. Consider the

The same i s s u e i s p a r t l y c o n t r o l l i n g a n o t h e r question: what ' p r i m i t i v e s ' d o w e use i n o u r representation? Some p r i m i t i v e s a r e i m p o r t a n t because o f t h e i n f e r e n c e s t h e y g e n e r a t e o r r e q u i r e . F o r e x a m p l e , i n m o d e l l i n g t h e e d u c a t i o n a l o r any i n s t i t u t i o n a l w o r l d , t h e e v e n t o f ASSIGNing v a l u e s to certain c h a r a c t e r i s t i c s is treated as a p r i m i t i v e because i t i m p l i e s t h a t t h e a g e n t o f t h e a c t i o n i s s o c i a l l y r e s p o n s i b l e f o r t h e a c t and has the authority to carry it out (e.g. only a profess o r i s a l l o w e d t o a s s i g n grades t o a s t u d e n t ) . O t h e r ' p r i m i t i v e s ' a r e n o t h i n g more t h a n a r b i t r a r y u n d e f i n e d terms ( e . g . WRITE), w h i c h a r e n o t i m p o r t a n t t o t h e p a r t i c u l a r domain o f d i s c o u r s e f o r which t h i s r e p r e s e n t a t i o n i s geared. Similarly, some e v e n t s w i l l b e d e s c r i b e d b y s c e n a r i o s i n t e r m s o f o t h e r e v e n t s o n t h e s e m a n t i c n e t ( e . g . APPLY i n s e c t i o n 2 . 3 . 1 ) w h i l e o t h e r s are assumed t o have a n e x t e r n a l d e f i n i n g s c e n a r i o w h i c h i s known t o e x i s t but is not e x p l i c i t l y stored (e.g. "completing an application form"). 2.

t h a t i t i s necessary t o discuss the v a r i o u s i n f e r e n c e s t h a t can be made f r o m edges in t h e n e t ( c a s e s , * ' s u b " , " p a r t " ) and s c e n a r i o s ( i n c l u d i n g c a u s a l and t e m p o r a l c o n n e c t i v e s ) . Many o f t h e i n f e r e n c e s resemble those discussed i n Rieger [ 1 2 ] . Since we are u s i n g a d i f f e r e n t r e p r e s e n t a t i o n than h i s , w e have f o u n d i t n e c e s s a r y t o r e s t a t e t h o s e i n f e r e n c e s i n our t e r m i n o l o g y . In a d d i t i o n , the use o f s c e n a r i o s w i t h i n t h e n e t w i l l , h o p e f u l l y , guide the i n f e r e n t i a l g e n e r a t i o n o f context t o p o t e n t i a l l y r e l e v a n t i n f o r m a t i o n , a s opposed t o b r u t e " e x p a n s i o n " as done by R i e g e r .

Generation of Context

Our s t u d y o f c o n t e x t h e r e i s c e n t e r e d a r o u n d i n f e r e n t i a l l y g e n e r a t i n g t h e " f o r e g r o u n d " (Chafe ( 1 1 ) for a given sentence. When r e g a r d i n g t h i s p r o b l e m f r o m a network p o i n t o f v i e w , one sees

When A i s i n s t a n t i a t e d , t h e r e i s a s t r o n g i n f e r e n c e t h a t B w i l l b e r e l e v a n t and t h u s i t s h o u l d b e i n s t a n t i a t e d and p l a c e d i n t h e c o n t e x t .

138

T h i s e x p e c t a t i o n e n a b l e s t h e system t o p r e d i c t p a r t of a scenario from a piece of i t . The one s t e p p r e d i c t i o n corresponds Sieger's resultative class. S i m i l a r l y , when B i s i n s t a n t i a t e d , e s p e c i a l l y when i t i s w i t h i n a w e l l s p e c i f i e d s c e n a r i o , t h e p r e s u m p t i o n t h a t A e x i s t s (and caused B ) i s s t r o n g . This class of inferences is s i m i l a r to Rieger's causative class. A p a r t i c u l a r l y i m p o r t a n t i n f e r e n c e t o b e made f r o m t h e " e f f e c t " edge i s t h a t o f s o c i a l o b l i g a t i o n . For example,

I f t h e " e f f e c t " edge c o n n e c t s two e v e n t s w i t h d i f f e r e n t a g e n t s , t h e n t h e agent ( s u p p l i e r ) o f t h e caused e v e n t ( s u p p l y ) i s under s o c i a l o b l i g a t i o n to perform the event or s c e n a r i o . We f e e l t h i s usage o f " e f f e c t " t o b e v e r y i m p o r t a n t t o i n s t i t u t i o n a l s c e n a r i o s and i t may b e a d i s t i n g u i s h i n g f e a t u r e of these scenarios from say, p h y s i c a l or mental s c e n a r i o s .

B c o u l d be a c h a r a c t e r i s t i c or an e v e n t . If A is i n s t a n t i a t e d , we can c o n f i d e n t l y a s s e r t B i n t o t h e c o n t e x t s i n c e A c o u l d n o t have happened u n l e s s B were s a t i s f i e d ( R i e g e r ' s " e n a b l i n g " c l a s s ) . I f B e x i s t s , w e can p r e d i c t t h a t A s h o u l d b e p a r t of the context. A s i m i l a r a n a l y s i s can b e c a r r i e d o u t f o r o t h e r edge l a b e l s such a s " p a r t " , " a g e n t " , "characteristic", "object", etc. With a l l edges, p r o p a g a t i o n i n the r e v e r s e d i r e c t i o n from the t a r g e t node i s dependent upon t h e number o f i d e n t i c a l l y l a b e l l e d edges e n t e r i n g t h a t n o d e .

Suppose now t h a t t h e i n p u t s e n t e n c e had been " J o h n S m i t h ' s a d d r e s s i s 6 5 S t . George S t . , T o r o n t o " , where i t i s n o l o n g e r s p e c i f i e d w h e t h e r we a r e t a l k i n g a b o u t a home or o f f i c e a d d r e s s . There a r e t h r e e choices a v a i l a b l e : ( i ) f i t the i n f o r m a t i o n below a l l s c e n a r i o s w e c a n , t h u s t r e a t i n g "65 S t . George S t . , T o r o n t o " b o t h a s a n o f f i c e and home a d d r e s s , ( i i ) f i t t h e new i n f o r m a t i o n below the f i r s t s c e n a r i o found o r ( i i i ) r e a l i z e the ambiguity, but not f i t t i n g the i n p u t g r a p h t o e i t h e r s c e n a r i o , and a s k i n g f o r c l a r i f i cation. We w i l l adopt s t r a t e g y ( i i i ) , a l t h o u g h there i s n o overwhelming evidence e i t h e r f o r i t o r the o t h e r s . I n g e n e r a l , graph f i t t i n g i s c a r r i e d o u t b y c o n s i d e r i n g a c a n d i d a t e s c e n a r i o , c a l l i t S , and a s t r u c t u r e s , and b y t r y i n g t o match e v e r y e v e n t o r c h a r a c t e r i s t i c in S to a corresponding e n t i t y in s. Once i t has been d e c i d e d t h a t s matches S , S i s added t o t h e ' m a t c h e d - s c e n a r i o ' l i s t . The SUBscenarios o f S a r e c o n s i d e r e d n e x t and each i s matched a g a i n s t s . A l l of t h e s e , which successf u l l y match a g a i n s t s , t h e n r e p l a c e S o n t h e matched-scenario l i s t . When t h i s s t e p n o l o n g e r p r o d u c e s any c h a n g e s , a s e a r c h i s c a r r i e d o u t " d o w n s t a i r s " , below e v e r y s c e n a r i o o n t h e m a t c h e d s c e n a r i o l i s t , t o see i f ( p a r t i a l ) i n s t a n t i a t i o n s o f s can b e f o u n d . Depending o n t h e r e s u l t o f t h i s search, either (a p o r t i o n of) s is i d e n t i f i e d w i t h a l r e a d y e x i s t i n g nodes o r new ones a r e c r e a t e d and p l a c e d b e l o w a l l s c e n a r i o s o n t h e l i s t . The f i r s t s c e n a r i o s t o b e c o n s i d e r e d f o r m a t c h i n g a g a i n s t s depend o n t h e c o n t e x t . Thus, i f S E N D - a - r e c o m m e n d a t i o n - l e t t e r i s o n t h e shadow d a t a base and a " s e n d " e v e n t appears a s p a r t o f s , g r a p h f i t t i n g w i l l f i r s t t r y t h e SEND w h i c h i s p a r t o f t h e c o n t e x t r a t h e r t h a n t h e g e n e r i c node SEND. I f t h i s match i s s u c c e s s f u l , e v e r y t h i n g proceeds a s d e s c r i b e d b e f o r e . If n o t , the system has been ' s u r p r i s e d ' b u t , r a t h e r t h a n g i v i n g u p , it t r i e s to f i n d scenarios higher up on the n e t , w h i c h w i l l match s . As explained in section 1 . 3 . 1 , t h i s s e a r c h w i l l always b e s u c c e s s f u l because o f t h e way t h e t o p o f t h e SUB a x i s i s organized. Of c o u r s e , once a p o s i t i o n f o r s has been f o u n d , a ( p o s s i b l y new) c o n t e x t i s g e n e r a t e d . Thus g r a p h f i t t i n g and c o n t e x t a r e i n t i m a t e l y r e l a t e d , a l t h o u g h t h e y have d i f f e r e n t f u n c t i o n s .

We have n o t y e t s t a t e d where t h e s e p r e d i c t i o n s should be placed w i t h respect to the r e s t of the net. We v i e w t h e c o n t e x t as a shadow d a t a base, one w h i c h can b e seen i f l o o k e d f o r b u t i g n o r e d when n e c e s s a r y . 2.2

I n a t t e m p t i n g t o i n t e g r a t e t h i s t o t h e semant i c n e t , w e c o u l d f i t " a d d r e s s " below t h e g e n e r i c ADDRESS n o d e , " t i m e " below t h e g e n e r i c TIME n o d e , etc. However, s i n c e w e w i s h t o i s o l a t e t h e most s p e c i f i c information in the net relevant to a g i v e n i n p u t , i t i s advantageous t o f i t t h e above example b e l o w t h e " c u r r e n t o f f i c e a d d r e s s " s c e n a r i o . T h u s , even t h o u g h " a d d r e s s " may o r i g i n a l l y match t h e g e n e r i c ADDRESS n o d e , t h e g r a p h f i t t i n g a l g o r i t h m w i l l push i t down t o more s p e c i f i c scenarios u n t i l t h i s is no longer p o s s i b l e . This i s t h e essence o f g r a p h f i t t i n g .

Graph F i t t i n g

A p a r t o f t h e p r o c e d u r e f o r i n t e g r a t i n g new i n p u t t o t h e s e m a n t i c n e t w o r k i s done b y t h e g r a p h f i t t i n g algorithm. This s e c t i o n discusses both t h e a l g o r i t h m and i t s r e l a t i o n t o c o n t e x t . Consider the sentence "John Smith's o f f i c e address i s 6 5 S t . George S t . , T o r o n t o " , w h i c h i s mapped i n a s t r a i g h t f o r w a r d manner i n t o

G r a p h - f i t t i n g can b e h e l p f u l i n i d e n t i f y i n g t h e head nouns m o d i f i e d b y r e s t r i c t i v e r e l a t i v e clauses. T h u s , f o r "The woman who s u p e r v i s e s J o h n s u p o r t s B i l l " , the a l g o r i t h m would f i r s t f i t the r e l a t i v e clause f o r the given i n f o r m a t i o n , l o c a t e t h e a p p r o p r i a t e woman, and p r o c e e d t o f i t t h e m a i n clause. In the process, the a l g o r i t h m would e n s u r e t h a t t h e woman i n q u e s t i o n i s a p r o f e s s o r .

139

2.3

An Example

Scenario D -

"PROCESS-an-application"

We now g i v e an example to i l l u s t r a t e t h e mechanisms d e s c r i b e d e a r l i e r . The example c o n s i s t s o f f o u r sentences t h a t are c o n s e c u t i v e l y p r e s e n t e d t o t h e s y s t e m and must b e " u n d e r s t o o d " . The sentences a r e : (1) (2) (3) (4)

2.3.1

"John a p p l i e d to the U n i v e r s i t y of Toronto" "He s e n t h i s a p p l i c a t i o n l a s t F r i d a y " " P r o f e s s o r Jones r e c e i v e d i t o n Monday" "John r e c e i v e d h i s acceptance on Thursday"

Scenarios of

the e d u c a t i o n a l w o r l d

The s c e n a r i o s t o b e used f o r t h e a n a l y s i s t h e f i v e s e n t e n c e s w i l l now b e d e s c r i b e d . Scenario A -

of

"ENTER-a-university"

I n o r d e r t o e n t e r a u n i v e r s i t y one must a p p l y f o r a d m i s s i o n i n t o one o f t h e programs o f f e r e d b y the u n i v e r s i t y . Once t h i s i s d o n e , t h e u n i v e r s i t y p r o c e s s e s t h e a p p l i c a t i o n and i f i t f i n d s i t a c c e p t a b l e , the a p p l i c a n t i s e l i g i b l e t o r e g i s t e r f o r the program t h a t he a p p l i e d f o r . 2.3.2

Analyzing

the

sentences

I t w i l l b e assumed t h a t t h e s y s t e m does expect the t o p i c of d i s c o u r s e to be about s t u d e n t s , applications to u n i v e r s i t i e s e t c . , so the scenario A w i l l become i m m e d i a t e l y r e l e v a n t w h e n t h e f i r s t sentence is presented f o r a n a l y s i s . Sentence

1:

"John a p p l i e d Toronto"

to

the

University of

The s e n t e n c e i s f i t t e d b e l o w t h e APPLY e v e n t of s c e n a r i o A. The r e l e v a n c e o f s c e n a r i o A c a u s e s a number o f nodes t o b e added t o t h e shadow d a t a base (see s e c t i o n 2 . 1 ) .

T h i s s c e n a r i o i n c l u d e s a m o r e g e n e r i c PROCESS e v e n t w h e r e t h e " r e s u l t " may b e " a c c e p t a n c e " o r " r e j e c t i o n " , and a SUBevent where t h e " r e s u l t " case i s o n l y " a c c e p t a n c e " . This representation accounts f o r the c o n d i t i o n a l i n the s c e n a r i o ( i f t h e u n i v e r s i t y a c c e p t s h i m , h e can r e g i s t e r ) .

Scenario B - "APPLY~to-a-university" In order f o r a PERSON to APPLY to a UNIVERSITY, w i t h the r e s u l t being an APPLICATION, he must prepare the a p p l i c a t i o n and have it t r a n s f e r r e d to the UNIVERSITY.

Nodes i n s m a l l l e t t e r s i n d i c a t e h e r e i n s t a n t i a t e d o b j e c t s o f t h e c o n t e x t , whereas those i n c a p i t a l l e t t e r s i n d i c a t e o b j e c t s t h a t appear t o b e r e l e v a n t b u t have n o t been i n s t a n t i a t e d y e t . Sentence

2:

"He

sent

the a p p l i c a t i o n

We w i l l ignore the temporal meaning o f t h i s s e n t e n c e .

last

Friday"

component o f t h e

F i r s t o f a l l , t h e r e are two problems o f r e f e r e n c e t h a t must be r e s o l v e d h e r e . "He" can be a s s o c i a t e d w i t h " J o h n " e a s i l y enough s i n c e t h e l a t t e r i s t h e o n l y c a n d i d a t e w h i c h meets t h e selectional r e s t r i c t i o n s for "He". "The a p p l i c a t i o n " i s r e s o l v e d b y u s i n g t h e shadow d a t a b a s e generated so far. I t m u s t now b e d e t e r m i n e d w h e r e " s e n d a n a p p l i c a t i o n " f i t s a n d how i t i s r e l a t e d t o t h e current context. S i n c e n o SEND a p p e a r s o n s c e n a r i o A, a s e a r c h is c a r r i e d o u t above and 140

below APPLY on t h e DEF a x i s , to l o c a t e an o c c u r e n c e o f SEND w h i c h i s r e l e v a n t t o t h e c u r r e n t c o n t e x t . T h i s s e a r c h succeeds wo l e v e l s below APPLY

However, t h e " o b j e c t " case o f RECEIVE i s p r e d i c t e d t o b e LETTER, n o t " a c c e p t a n c e " . In a t t e m p t i n g to e x p l a i n t h i s c o n t r a d i c t i o n , we note t h a t both e n t i t i e s a r e used i n cases o f WRITE, w h i c h i s a c a n d i d a t e f o r t h e shadow d a t a b a s e , LETTER b e i n g t h e " r e s u l t " and " a c c e p t a n c e " b e i n g t h e " t o p i c " o f WRITE. Thus " a c c e p t a n c e " i s t h e message w h i l e LETTER i s t h e medium, and t h e s e terms a r e o f t e n used i n t e r c h a n g e a b l y ( " I r e c e i v e d ( t h e l e t t e r w i t h ) the message.", " I understood {the contents o f ) the book.". W e can f o r m a l i z e t h i s r u l e b y s a y i n g t h a t e n t i t i e s w h i c h s e r v e a s " t o p i c " and " r e s u l t " o f a n e v e n t i n v o l v i n g t r a n s m i s s i o n of a message may appear i n t e r c h a n g e a b l y i n a s u r f a c e s t r u c t u r e . T h i s e n a b l e s t h e s y s t e m t o a c c e p t t h e RECEIVE i t has f o u n d and to i n s t a n t i a t e PROCESS, EVALUATE, COMMUNICATE, WRITE, TRANSFER, SEND and RECEIVE in t h e shadow d a t a base because o f t h e " f e d " and " p r e r e q " edges t h a t c o n n e c t them t o t h e i n s t a n tiated receive.

and we can t h e r e f o r e c o n c l u d e t h a t s c e n a r i o s B and C are a l s o r e l e v a n t at t h i s p o i n t . The expanded shadow d a t a base i n c l u d e s t h e nodes shown e a r l i e r plus :

N o t e t h a t " a p p l i c a t i o n " h a s now b e e n i n s t a n t i a t e d along w i t h " c r e a t e " , which is a p r e r e q u i s i t e for TRANSFER, a n d t h e r e f o r e " s e n d " . Sentence

3:

"Professor Monday"

Jones

received

it

on

A l t h o u g h n o t shown, t h e r e s u l t i n g shadow d a t a base i s f a i r l y l a r g e b y now and i n c l u d e s s e v e r a l i t e m s w h i c h were i n f e r r e d a n d / o r p r e d i c t e d b e c a u s e of the scenarios t h a t are a v a i l a b l e to the system.

T h e r e i s a RECEIVE n o d e t h a t i s p a r t o f t h e c o n t e x t and i t b i n d s " i t " t o " a p p l i c a n t " , a s i t should. U n f o r t u n a t e l y , i t has a s " s o u r c e " t h e " u n i v e r s i t y o f t o r o n t o " , whereas the source o f the incoming " r e c e i v e " is "professor jones". To r e s o l v e t h i s p r o b l e m , w e must a c e r t a i n t h a t " p r o f e s s o r j o n e s " is employed by the U n i v e r s i t y , has t h e a u t h o r i t y t o r e v i e w a p p l i c a t i o n s and i s i n the r i g h t department. W e assume t h a t t h e s e c h e c k s can b e c a r r i e d o u t and t h e y s u c c e e d , s o t h a t t h e RECEIVE n o d e o f t h e shadow d a t a b a s e c a n b e instantiated.

3.

We have e x t e n d e d s t a n d a r d s e m a n t i c n e t w o r k methodology t o i n c l u d e s c e n a r i o s i n o r d e r t o d e a l w i t h p r o b l e m s o f i n f e r e n c e and c o n t e x t . Along w i t h t h e d e c l a r a t i v e knowledge s t o r e d o n a s e m a n t i c n e t , w e have s k e t c h e d c l a s s e s o f p r o c e d u r a l knowledge t h a t can b e used t o r e p r e s e n t p e r i p h e r a l k n o w l e d g e , t o map f r o m one l e v e l o f t h e r e p r e s e n t a t i o n t o a n o t h e r , and t o e x e c u t e i m p e r a t i v e s where t r a c i n g through scenarios would be too expensive. Like o t h e r workers proposing s i m i l a r s t r u c t u r e s , we c o n s i d e r s c e n a r i o s as o r g a n i z e d chunks of knowledge w h i c h model commonplace o c c u r e n c e s i n our e d u c a t i o n a l w o r l d . I t i s a p p a r e n t t h a t more i n s t i t u t i o n a l w o r l d s w i l l have t o b e m o d e l l e d b e f o r e w e can d e t e c t c o m m o n a l i t i e s and can make d e f i n i t i v e s t a t e m e n t s o n t h e k i n d s o f p r i m i t i v e s needed t o model such worlds. A n o t h e r v e r y i m p o r t a n t p r o b l e m w e have n o t d i s c u s s e d a t a l l i n t h i s paper i s t h a t o f switching topics. Any r e a s o n a b l e language u n d e r s t a n d i n g system must have a model ( s c e n a r i o ? ) of how, when and why t h e t o p i c under d i s c u s s i o n switches. M o r e o v e r , a d d i t i o n a l work i s needed t o d e t e r m i n e t h e amount o f s c e n a r i o i n f o r m a t i o n w h i c h must b e b r o u g h t i n t o a c o n t e x t when a n e v e n t i n the net i s i n s t a n t i a t e d . I n a d d i t i o n t o these p r o b l e m s , w e have n o t d e a l t w i t h e x c e p t i o n s t o s c e n a r i o s , i . e . , what happens when t h i n g s d o n ' t o c c u r a s t h e y are supposed t o .

Note t h a t " t r a n s f e r " has a l s o been i n s t a n t i a t e d s i n c e i t s s c e n a r i o has been i n s t a n t i a t e d completely. The e v e n t " a s s i g n . a u t h o r i t y " i s used a s a simple-minded representation of the fact t h a t P r o f e s s o r Jones has t h e a u t h o r i t y t o h a n d l e (and process) applications. Sentence

4:

"John received h i s Thursday"

acceptance

C o n c l u d i n g Remarks

on

A g a i n , i g n o r i n g the temporal m o d i f i e r , the s t r u c t u r e t h a t w e s t a r t w i t h and w h i c h must b e i n t e g r a t e d to the semantic net is

" A c c e p t a n c e " i s a n i n s t a n t i a t i o n o f a node a l r e a d y o n t h e shadow d a t a base and t h i s i s used t o g u i d e t h e s e a r c h t h a t w i l l e x p l a i n " r e c e i v e " above and b e l o w PROCESS.

Among a l l t h e s e and o t h e r open problems we d i d not mention above, we consider the problem of m o d e l l i n g s e v e r a l i n s t i t u t i o n a l w o r l d s and t h a t o f m o d e l l i n g the conversation process, i n a d d i t i o n t o knowledge a b o u t w h i c h t o c o n v e r s e , o f paramount

141

importance. We i n t e n d to work on t h e s e p r o b l e m s i n the f u t u r e . Acknowledgements We would l i k e to thank Nick Roussopoulos, H a r r y Wong, John T s o t s o s and C o r o t Reason f o r s e v e r a l u s e f u l d i s c u s s i o n s t h a t have h e l p e d u s f o r m u l a t e the ideas expressed i n t h i s paper. The r e s e a r c h d e s c r i b e d i n t h i s paper was p a r t i a l l y s u p p o r t e d b y t h e Department o f Communic a t i o n s o f Canada and b y t h e N a t i o n a l Research C o u n c i l o f Canada. References 1.

2. 3.

4. 5.

6.

7.

8.

9.

10.

11.

12.

Minsky, M. " A Framework f o r t h e R e p r e s e n t a t i o n o f Knowledge", i n Psychology o f V i s i o n , W i n s t o n , P . ( E d . ) , McGraw H i l l , 1975. Schank, R. and A b e l s o n , R. " S c r i p t s , Plans and K n o w l e d g e " , P r o c e e d i n g s 4 I J C A I . Abelson, R. "The S t r u c t u r e o f B e l i e f S y s t e m s " , i n Computer Models o f Thought and Language, Schank and C o l b y ( E d s . ) , W.H. Freeman and C o . , 1 9 7 3 , p p . 2 8 7 - 3 3 9 . Norman, D . and R u m e l h a r t , D . E x p l o r a t i o n s in C o g n i t i o n , W.H. Freeman, 1975. M a r t i n , W. A u t o m a t i c Programming Group Memos 6 , 8 , 1 1 , 1 2 , 13; a l s o OWL Memos 1 , 2 , 3 , 4 , MIT. Malhotra, A. "Design C r i t e r i a f o r KnowledgeBased E n g l i s h Language Systems f o r Management: A n E x p e r i m e n t a l A n a l y s i s " , Ph.D. t h e s i s , P r o j e c t MAC, MIT, 1975. M y l o p o u l o s , J . , B o r g i d a , A . Cohen, P . , R o u s s o p o u l o s , N . , T s o t s o s , J . , Wong, H . "TORUS—A N a t u r a l Language U n d e r s t a n d i n g System f o r D a t a Management", P r o c e e d i n g s 4IJCAI. Q u i l l i a n , R. "The Teachable Language Comprehender", CACM, 12 ( 8 ) , 1969, p p . 459-476. Simmons, R. "Semantic Networks: Their C o m p u t a t i o n and Use f o r U n d e r s t a n d i n g E n g l i s h S e n t e n c e s " , i n Computer Models o f Thought and Language. Riesbeck, C. "Computational Understanding A n a l y s i s o f Sentences and C o n t e x t " , P h . D . T h e s i s , S t a n f o r d U n i v e r s i t y , 1974. C h a f e , W. " D i s c o u r s e S t r u c t u r e s and Human K n o w l e d g e " , i n Language Comprehension and t h e A c q u i s i t i o n o f K n o w l e d g e , F r e e d l e and C a r o l l ( E d s . ) , W i n s t o n , 1972. Rieger, c " C o n c e p t u a l Memory: A Theory and Computer Program f o r P r o c e s s i n g t h e Meaning C o n t e n t o f N a t u r a l Language U t t e r a n c e s " , P h . D . t h e s i s , S t a n f o r d A I Memo 233, S t a n f o r d U n i v e r s i t y , 1974.