A formal model for fuzzy flexible querying over deductive databases with uncertainty

Peter Vojtáš

Katedra matematickej informatiky, Univerzita P. J. Šafárika
Jesenná 5, 041 54 Košice, Slovak Republic
vojtas@kosice.upjs.sk

Abstract. We consider flexible querying of crisp extensional data. Flexibility deals with vagueness of the natural language, uncertainty of the background knowledge and similarity on attribute domains. We discuss several motivating examples and overview different approaches to solution. We present a model based on fuzzy logic in narrow sense with multiple connectives and aggregations. We hope that this rich language can better describe real world situations. We construct a corresponding tuple based fuzzy relational algebra for query computation. A sound and complete theory of fuzzy datalog with max-product similarity is based on the existence of a computable fixpoint.

Keywords: Natural language queries, user preferences and stereotypes, context of the query, uncertainty, approximate reasoning, deductive databases, knowledge based systems, vagueness, similarity, integration of information retrieval and database systems, intended semantics, weighted scoring rules, flexible querying.

<< Obsah