Relational database reverse engineering and terminological reasoning

Abstract

Reverse engineering of relational databases has known an increasing interest due to the need to have a more effective use of databases whose semantics has become uncertain. For a reverse engineering to be efficient, realistic assumptions have to be made on the physical DB schema and the degree of involvement of the designer for providing knowledege on data and for validating results. This paper describes a method which exploits various sources for gathering information on data and proposes validation tools based on the use of terminological logics. Moreover no assumption on physical schemas is needed (unless 1rst normal form).