Syntactic and semantic knowledge in the DELPHI unification grammar

Abstract

This paper presents recent natural language work on HARC, the BBN Spoken Language System. The HARC system incorporates the Byblos system [6] as its speech recognition component and the natural language system Delphi, which consists of a bottom-up parser paired with an integrated syntax/semantics unification grammar, a discourse module, and a database question-answering backend. The paper focuses on the syntactic and semantic analyses made in the grammar.

Computational aspects of discourse in the context of MUC-3

“…The discourse module of PLUM is new (as are all the other PLUM components except the parser), an d in flux-what is reported here is its state as of the MUC-3 test . This module is a significant departure from the discourse module in BBN ' s Janus [3] and the related module in BBN ' s Delphi [4,5] . The discourse components of those systems, which were developed primarily for question-answering applications in limited domains, are able to take advantage of having complete analyses of the input sentences .…”

A relaxation method for understanding spontaneous speech utterances

“…Current approaches to the language understanding aspect of spoken language systems tend to fall into two categories. In syntax-driven formulations [1,4,10], a complete syntactic analysis is performed which attempts to account for all words in an utterance. While providing strong linguistic constraints to the speech recognition component and a useful structure for further linguistic analysis, such an approach can break down in the presence of unknown words, novel linguistic constructs, recognition errors, and some spontaneous speech events such as false starts.…”

The mapping unit approach to subcategorization

“…Such approaches as Definite Clause Grammar [11], Categorial Grammar [1], PATRqI [13], and lexicalized TAG [12] inelude in one form or another a notion of "subcategorization frame" that specifies a sequence of complement phrases and constraints on them. Some have also advocated using the feature system to encode semantic information (as for example [9]), and this has recently characterized our own approach [3]. "Mapping unit" subcategorization is partly inspired by these approaches, but it handles several kinds of variation in natural language utterances which cause difficulty for them.…”