Deriving rules for medical expert systems using natural language parsing and discourse analysis
F. J. Finaldo · 1989
A rule knowledge base for a rule-based expert system is generated automatically by processing input text in the form of published papers. This system 'reads' the input text and using sublanguage analysis techniques performs syntactic, semantic and discourse analysis of the text. The syntactic analysis phase uses the Linguistic String Parser developed at New York University. Some of the semantics are integrated in with the LSP. The final semantic interpretation and discourse analysis is determined by a computer program written in C. The information derived is stored internally in a frame-like structure which is used to resolve ambiguity and determine the proper context of the phrase and/or sentence being analyzed. This internal structure is then traversed with the assistance of a medical knowledge base to generate production rules for the expert system.