Conceptual Analysis of Noun Groups in English

An expectation-based system, NGP, f o r parsing Engl ish noun groups i n to the Conceptual Dependency representat ion is descr ibed. The system is a par t of ELI (Engl ish Language I n t e rp re te r ) which is used as the f ron t end to several na tu ra l language understanding systems and is capable of handl ing a wide range of sentences of considerable complex i ty . NGP processes the input from l e f t to r i g h t , one word at a t ime, using l i n g u i s t i c and world knowledge to f i nd the meaning of a noun group. D ic t ionary en t r i es fo r i n d i v i d u a l words contain much of the program's knowledge. In a d d i t i o n , a l i m i t e d a b i l i t y f o r the handl ing of s l i g h t l y i nco r rec t sentences and unknown words is incorpora ted. 0. I n t roduc t i on Every na tu ra l language processor has to have the a b i l i t y to i n t e r p r e t noun phrases. This paper describes a set of programs ca l led NGP (Noun Group Processor) which is an i n t e g r a l par t of ELI , the Engl ish Language In te rp re te r (Riesbeck and Schank 1976) which serves as the f r o n t end to three of the Yale na tu ra l language understanding systems, SAM, PAM and WEIS. SAM is a system capable of understanding s to r i es such as var ious newspaper repor ts by using s c r i p t s (Schank and Abelson 1975, 1977; Cu l l i ng fo rd 1975, 1977). PAM is an understanding system which uses general knowledge about peoples' goals and plans (Wilensky 1976). WEIS is a system which understands and c l a s s i f i e s a great v a r i e t y of i so la ted newspaper headlines on i n t e r n a t i o n a l r e l a t i o n s . Thus, our task was to process not only noun phrases of considerable complexity but also to i n t e r p r e t newspaper headl ines, which are not always grammatical ly c o r r e c t . The fo l l ow ing two examples i l l u s t r a t e the kind of sentences our system is able to handle. 1. A CONNECTICUT MAN, JOHN DOE, AGE 23, OF 342 COLLEGE AVENUE, NEW HAVEN WAS PRONOUNCED DEAD AT THE SCENE BY DR. DANA BLAUCHARD, MEDICAL EXAMINER. 2. FUNERAL OF INDIA 'S SHASTRI ATTENDED BY USSR KOSYGIN AND USA HUMPHREY. This work was supported in par t by the Advanced Research Projects Agency of the Department of Defense and monitored under the Of f i ce of Naval Research under cont rac t N00014-75-C-1111 To process such a la rge scope of sentences the program makes extensive use of i t s knowledge of the problem domain and the redundancy of n a t u r a l language expressions. This saves e f f o r t and permits co r rec t processing of such i r r e g u l a r i t i e s of input t ex t as missing commas and a r t i c l e s , or s l i g h t l y i nco r rec t word order. I t also provides f o r the a b i l i t y to ignore unknown words or ( i n some cases) to make p laus ib le i n t e r p r e t a t i o n s of unknown words. This knowledge is kept in the d i c t i o n a r y . The con t ro l mechanisms remain domain independent. NGP is a p roduc t i on l i ke system which uses expectat ions as i t s basic con t ro l mechanism. The problem wi th every p roduc t i on l i ke system is the tendency f o r the accumulation of a la rge number of expectat ions f i g h t i n g fo r a chance to be tes ted . In t h i s work I have t r i e d to develop a theory of how var ious expectat ions are organized and processed, which, I be l i eve , is in f ac t a theory of how people process na tu ra l language. The basic gu id ing p r i n c i p l e fo r t h i s theory was i t s i n t u i t i v e p l a u s i b i l i t y . !• Noun Group Semantics In t h i s paper we w i l l discuss two classes of noun groups according to the conceptual s t r uc tures they generate: PP P ic tu re Producers and CTP Concept Producers. PP's are def ined by Schank (Schank 1975) as concepts which tend to produce p ic tu res of r e a l wor ld items in the mind of a hearer. For example,