Two discourse generators. A grammar and a lexicon for a text-production system
Computer generation of text, linguistically justified grammars, KDS, PROTEUS, natural language, systemic grammar, text generation. Computer generation of text linguistically justified grammars, knowledge representation methods, PENMAN, natural language, systemic grammar, text generation. Because discourse generation is a relatively new branch of Artificial Intelligence, only a few incomplete discourse generators have been developed. A study of those efforts reveals the nature of the task, what makes it difficult, and how the complexities of discourse generation can be controlled. This report compares two recent discourse generation systems: PROTEUS, by Anthony Davey at the University of Edinburgh, and KDS, by Mann and Moore at USC/Information Sciences Institute. Viewing the systems separately, the author identifies particular techniques in each system that contribute strongly to the quality of the resulting text. A comparison of the two systems follows, with a discussion of their common failings and the possibility of creating a new system that combines the strengths of both PROTEUS and KDS. In a text-production system high and special demands are placed on the grammar and the lexicon. This report views these components in such a system. First, the subcomponents dealing with semantic and syntactic information are presented separately. The problems of relating these two types of information are then identified. Finally, strategies designed to meet the problems are proposed and discussed. One of the issues illustrated is what happens when a systemic linguistic approach is combined with a KL-ONE-like knowledge representation--a novel and hitherto unexplored combination.
- Research Organization:
- Air Force Office of Scientific Research, Bolling AFB, DC (USA)
- OSTI ID:
- 6755353
- Report Number(s):
- AD-A-126353/2; AFOSR-TR-83-0113
- Country of Publication:
- United States
- Language:
- English
Similar Records
Design of a knowledge-based report generator
Towards a semantic lexicon for biological language processing