Modified caseframe parsing for speech understanding systems

Abstract

We suggest for speech understanding systems a esseframe paraing strategy which devistes from 'pure' casefrsme parsing in st lesst two respects: paraing is not exclusively bssed on top-down instantiation of casefrases, and casefrases are Merged before use with syntactic knowledge. Our strategy has been developed to be executed m& an inferential process in which a set of Knowledge Sources cooperate through the blackboard. The Knowledge Sources are automatically defined by Merging syntactic and semantic knowledge expressed declsrstivslyi this integration of ayntax rules and caseframes Makes it possible to exploit siMultsneously both syntactic and semantic constrsints