Semantic thesaurus for automatic expanded query in information retrieval

This paper proposes (a) a semantic structuring for thesauri and (b) a procedure that handles it, for automatic query expansion in information retrieval. The thesaurus for this experiment was built manually, based on a traditional dictionary, adopting aspects from the Generative Lexicon Theory by James Pustejovsky as well as concepts from object oriented software modeling. We show how to select new terms for query expansion and to calculate their weights. This last task is performed according to intersections of the derived lexical sets and to the depth level for descriptors search with respect to each considered term. Also, an evaluation of the use of these resources is presented.