A Knowledge-Based Approach For Database Flexible Querying

People have often insufficient knowledge about database structure and contents, thus frequently obtaining empty answers or having to reformulate the queries several times. This paper describes an approach extending the search criteria ranges in order to provide approximate answers to the user. This approach takes into account the user preferences. It is based on a knowledge base (KB) related to the application domain. The initial SQL query is transformed into a flexible query based on KB and fuzzy set theory. This flexible query is then rewritten into a Boolean query called envelope. The envelope is evaluated with a traditional DBMS to take advantage of its full optimization capabilities. The tuples satisfying the envelope are finally ranked according to their satisfaction degree. Unlike other approaches, this one requires to modify neither SQL language nor the database engine