Reverse Engineering of Relational Databases to Ontologies

A majority of the work on reverse engineering has been done on extracting entity-relationship and object models from relational databases. There exist only a few approaches that consider ontologies, as the target for reverse engineering. Moreover, the existing approaches can extract only a small subset of semantics embedded within a relational database, or they can require much user interaction for semantic annotation. In our opinion, the potential source of these problems lies in that the primary focus has been on analyzing key correlation. Data and attribute correlations are considered rarely and thus, have received little or no analysis. As an attempt to resolve the problems, we propose a novel approach, which is based on an analysis of key, data and attribute correlations, as well as their combination. Our approach can be applied to migrating data-intensive Web pages, which are usually based on relational databases, into the ontology-based Semantic Web.