The authors propose a system for incremental construction of a knowledge base (KB) based on terminological logic (TL) subsumption criteria. The information to be introduced in the KB corresponds to a set of analysed assertions which are produced by a syntactic and semantic parser. The assertions correspond to a particular type of natural language sentences related to the medical field. Each parsed sentence corresponds to a predicate calculus (PC) formula whose variables are annotated by TL expressions. Each formula is stored in the KB according to a set of TL subsumption criteria that can be used, for example, to avoid the introduction of redundancies and contradictions in the KB. These criteria allow us to simplify the KB and also to simplify the proof tree during information retrieval
Published in:
Systems, Man, and Cybernetics, 1998. 1998 IEEE International Conference on
(Volume:2
)
Date of Conference: 11-14 Oct 1998