Bayesian Discriminant Analysis for Lexical Semantic Tagging

Nicolas Turenne (1, 2)

1 INRAE - Institut National de Recherche pour l’Agriculture, l’Alimentation et l’Environnement
2 LIAB - Laboratoire d'intelligence artificielle et biométrie

Structuring of terminology automatically is interesting for information extraction or indexing. It remains to evaluate the results obtained. We use a clustering method to build term classes assuming that (even) an incomplete thesaurus could inform the user about the semantic interpretation of classes. A variant naive Bayesian analysis is used to model an association network between terms and categories. Conditional relations express a causal membership of a term to several categories. We show that a lexical tagging of terms classes, obtained from corpora processing, performs with a high degree of probability

Domaines: Informatique [cs]

Nicolas Turenne. Bayesian Discriminant Analysis for Lexical Semantic Tagging. European Meeting on Cybernetics and Systems Research (EMCSR), Apr 2002, Vienne, Austria. ⟨hal-03373905⟩