Research And Implementation Of Ontology Extraction Based On Relational Database

Ontology is a conceptual model which is capable of describing information system at the semantic and knowledgeable level. It is primarily used to organize, express and manage knowledge. Ontology is already widely used in lots of information system fields such as knowledge management, information retrieval, semantic web, e-commerce and intellectual information system integrations. In order to make good use of ontology model's advantages, greatly improve share, exchange and re-usage in information systems, the ontology creation of existed knowledge system is the first problem that needs to be solved. Nevertheless, ontology's creation is a very complicated and tedious job. Current ontologies are mostly manually created by experts, which severely hampered the efficiency of ontology creation, and therefore make the ontology development time cycle longer. Because of this, this thesis treats the relational database as data source, proposes an approach to automatically extract ontologies from relational database, so that decrease the ontology development's time cycle and enhance the efficiency of ontology creation.This thesis firstly analyzes relational schema and owl ontology schema, proposes a projection rule between relational schema and owl ontology schema and creates the initial owl ontology according to the rule, then proposes a comprehensive similarity calculation method which combines the structural similarity computed by edit distance and a semantic similarity computed by WordNet similarity, and therefore conduct ontology integration based on concept similarity. Meanwhile, thesis also proposes a serial of WordNet based approaches to expand initial ontology's conceptual hierarchy and equal concepts, which improves the ontology extraction's practicability and effectiveness. Finally, this thesis implements a relational database based ontology creation system under the foundation of the researches in this thesis. The system successfully extracts ontologies from relational database, effectively reduces the time cost of ontology creation and improves the efficiency of ontology development.