SAGE: a sentence parsing and generation system

SAGE (Sentence Analysis and GEneration system) is an operational parsing and generating system. It is used as a Natural Language Frontend for Esprit project Esteam--316, whose purpose is to advise a novice user through a cooperative dialogue.The aim of our system is to validate the use of a Lexicon--Grammar (drawn from the LADL studies) for sentence--parsing and generation, and to implement linguistic knowledge in a declarative way using a formalism based upon Functional Descriptions (FD). We have also developed the parser and the generation module so that they share informations and knowledge bases as much as possible: they work on the same semantic dictionary and the same linguistic knowledge bases, except that they have their own grammar. We have also implemented a tracking of semantic objects that have been instantiated during a dialogue session: the so-called Token History is provided for semantic reference and anaphor resolution during parsing and for pronoun production during generation.After introducing to Esteam--316, this paper describes the linguistic knowledge bases required by SAGE, and then focuses on the Generation Module. Section 4 explains how pronouns are handled. The last section is a brief evaluation of our present work.