Learning Semantics from Relational Database Schema and Data

Abstract

Ontology has been proposed to resolve the issues of semantic heterogeneity in data interoperability and data sharing among heterogeneous systems because it provides an effective approach to represent the knowledge about data of a specific domain and has been applied in many fields, such as Semantic Web, e-commerce and information retrieval, etc. However, building ontology for information source manually is not only hard and error-prone, but also very personal if there is no common guideline. We propose a framework for data interoperability using semantic Web service to resolve the semantic heterogeneity in healthcare environment. We found that learning ontology from existing information resources is a good solution to explicitly express the semantics of information source, but some semantics may be missing during the ontology learning process. Since relational database is widely used for storing source data, in this paper a new approach of learning OWL ontology from a relational database is proposed. In order to acquire the complete conceptual information, a group of learning rules are used to obtain OWL ontology, including classes, properties, property characteristics, cardinalities and instances. The proposed learning rules are being implemented as a prototype. KeywordsOntology; Ontology Learning; OWL; Relational Database; Relational Model