[en] This publication was presenting another key aspect of the automatic analyser of French SABA: the introduction of “dual frames” as a new knowledge representation tool for expressing semantic dependencies between concepts.
Understanding natural language sentences requires to relate the concepts and relations expressed by them to the knowledge representation of a world model (or domain model), which is typically a kind of ontology, organizing concepts in hierarchical structures enriched with additional semantic relations. Dual frames were descriptions of the different types of semantic relationships that could be established between concepts occurring in a sentence. They could be assigned to all meaningful elements of a sentence and could be derived from more generic concepts by using ontology-based inheritance, and could be combined using intersection and union operations, to infer the semantic dependencies of complex sentence structures such as noun phrases and subordinate clauses. By matching the frames of a main term and its complements, semantic dependencies could be identified without syntactic analysis of the sentence, even in the presence of syntactic mistakes, thus making possible a robust natural language analysis.
Disciplines :
Computer science
Author, co-author :
Binot, Jean-Louis ; Université de Liège - ULiège > Dép. d'électric., électron. et informat. (Inst.Montefiore) > Réseaux informatiques
Ribbens, Daniel
Language :
English
Title :
Dual frames: a new tool for semantic parsing
Publication date :
August 1986
Event name :
Fifth American Conference on Artificial Intelligence (AAAI86)
Event date :
August 1986
Audience :
International
Main work title :
Proceedings of the fifth American Conference on Artificial Intelligence (AAAI86)