A Framework for Ontology-Based Top-K Global Schema Generation

Abstract

Global schema generation is the problem of generating a unified and merged schema based on existing heterogeneous local schemas and a set of correspondences which are generated by a schema-matching algorithm. The existing schema integration approaches only target the flat schema structure, but cannot handle hierarchical schema structure well, and may produce redundant information in the global schema due to the weak entities in local schemas. To deal with these kinds of problems, a framework for ontology-based top-k global schema generation is proposed in this paper. The framework employs ontology as a base-merging model to create the global schema with higher quality but less user involvement, because ontology can provide semantic and detailed constraints and develop ways to preserve the hierarchical structure and remove redundant entities. The proposed approach consists of three contiguous steps: (1) local schemas are converted to local ontologies; (2) local ontologies are merged and top-k global ontologies are generated; (3) the user chooses one merged global ontology, which is converted automatically back to the global schema. After comparing the proposed approach with an existing state-of-the-art schema integration approach, the proposed approach better preserves the hierarchical structure and removes the redundant information. As a result, the quality of the generated global schema is higher.