PHRAN - A Knowledge-Based Natural Language Understander

PHRAN - A Knowledge-Based Nature] Language Understender Robert Wilensky and Yigal Arena University of California at Berkeley Abstract We have developed an approach to natural language processing in which the natural language processor is viewed as a knowledge-based system whose knowledge is about the meanings of the utterances of its language. The approach is orzented around the phrase rather than the word as the basic unit. We believe that this paradi~ for language processing not only extends the capabilities of other natural language systems, but handles those tasks that previous systems could perform in e more systematic and extensible manner. We have construqted a natural language analysis program called PHRAN (PHRasal ANalyzer) based in this approach. This model has a number of advantages over existing systems, including the ability to understand a wider variety of language utterances, increased processlng speed in some cases, a clear separation of control structure from data structure, a knowledge base that could be shared by a language productxon mechanism, greater ease of extensibility, and the ability to store some useful forms of knowledge that cannot readily be added to other systems. 1.0 INTRODUCTION The problem of constructing a natural language ~rocessing system may be viewed as a problem oz constructing a knowledge-based system. From this orientation, the questions to ask are the following: What sort of knowledge does a system need about a language in order to understand the meaning of an utterance or to produce an utterance in that language? How can this knowledge about one's language best be represented, organized and utilized? Can these tasks be achieved so that the resulting system is easy to add to and modify? Moreover, can the system be made to emulate a human language user? Existing natural language processing systems vary considerably in the kinds of knowledge about language they possess, as well as in how thxs knowledge is represented, organized and utilized. However, most of these systems are based on ideas about language that do not come to grips with the fact that a natural, language processor neeos a great deal of knowledge aoout the meaning of its language's utterances. Part of the problem is that most current natural language systems assume that the meaning of a natural language utterance can be computed as a function of the constituents of the utterance. The basic constituents of utterances are assumed to be words, and all the knowledge the system has about ~he semantics of its language zs stored at the word level (~i~nbaum etal, 1979) (Riesbeck et al, 1975) (Wilks, 197~) (Woods, 1970). However, many natural language utterances have interpretations that cannot be found by examining their components. Idioms, canned phrases, lexical collocations, and structural formulas are instances of large classes of language utterances whose interpretation require knowledge about She entire phrase independent of its individual words (Becker, 19q5) (Mitchell, 19~71). We propose as an alternative a model of language use that comes from viewing language processing systems as knowledge-based systems tha£require the representation and organization of …

Content truncated to 500 words