Semantic Schema Mapping Using Property Precedence Relations

Information integration has always been a challenging issue in information management. Regardless of the implementation techniques, relations between heterogeneous data sources are stated in terms of high level specifications called schema mappings. Generally, such mappings are created using referential integrity constraints and a set of correspondences between properties that are created automatically by schema matchers or manually by users in a visual interface. Since different data sources are developed independently, schema mapping involves resolving the semantics of data and schema in different sources. In this paper, we study the fundamental semantic relations between properties from an ontological point of view, and show how one of the basic and natural relations between properties -- inferring implicit properties from existing properties -- is largely ignored. We discuss ontological foundations for such relations, and propose novel algorithms for schema mapping generation that exploit these relations. The proposed algorithms are capable of generating some new mappings that cannot be discovered in traditional methods.