Fuzzy Query Languages for Multimedia Data

Abstract

This chapter presents a fuzzy-based algebra, called SAMEW, to query multimedia objects. SAMEW allows for dealing within a common framework with several aspects relevant to similarity query processing as well as with the inherent imprecision that characterizes data, user requests and query results. Non-Boolean, namely fuzzy and similarity, predicates are used to rank tuples according to specific criteria. Complex multi-predicate queries can be formed by means of logical connectives, whose semantics is parameterized in order to adapt to specific scenarios. The same holds for the semantics of algebraic operators. These include properly extended traditional relational operators and new operators which allow threshold and best-matches queries to be easily expressed. A further important feature of SAMEW is the possibility of weighting both predicates and operands of algebraic operators so as to better fit user preferences/requirements. A working example dealing with Web data is used throughout the chapter to show the potentialities of SAMEW. Optimization issues are also briefly discussed.