Refining the thesaural associative relationship by applying the EuroWordnet semantic model.

Thesauri are tools which semantically organize a domain of knowledge for operational purposes. Their relational semantics is concerned with methods that connect terms with related meanings and are designed to support information retrieval, namely enhancing the information recall performance and contributing to improve precision, too. The network of relations of a thesaurus has a semantic function: by means of it, in fact, a representation of the meaning of each thesaurus term is provided, as well as of the conceptual structure of a subject area. The traditional thesaurus format - as described in international standards - includes three basic relationships, two of them at a conceptual level, hierarchical and associative relations, the other mostly at a lexical level, relation of equivalence. However, a rather widespread opinion is that this format should be refined in order to cope with the current needs of information organization. This refinement is necessary to enhance thesaurus suitability for uses in artificial intelligence (AI) and in the Semantic Web environments, as well as to increase possibilities for IR. This paper discusses the possibility of refining the associative relation into a number of sub-kinds by adopting the semantic model of EuroWordNet (EWN), as it was used, according to one of its national versions, ItalWordNet (IWN), to structure a terminological resource for a specific domain. A number of issues that such a work of refinement could imply, and in particular its domain dependence, are also discussed.