Relational Database-Oriented Ontology Automatic Construction

Semantic Web is the product of network era. With the rapid development of semantic web, ontology is widely used in industry. Ontology construction is the foundation of the research of ontology technology. The effectiveness and quality of ontology construction are two important issues which have direct impact on ontology applications. So far there are amount of researches about ontology construction approaches. However, some problems still need to be solved:Firstly, in the process of construction, resources in the existing relational database are not taken full advantages of. Secondly, due to the different level of expertise and their naming conventions, the ontologies in the same area are quite different so that they cannot be interoperated. Thirdly, traditional ontology integration requires the participation of experts in several areas. The integration results will be inconsistent because of the different views in the same area.In this paper, we propose our solutions to deal with the problems mentioned above. The relational database is utilized as the data source to process the data structure and records in relational database. Data structure and records are automatically converted to the corresponding ontology logical structures and ontology instances respectively. Furthermore, we develop the semi-finished ontology extraction system to solve the automatic extracting issue of ontology. To deal with the problem of interoperation, FC A (Formal Concept Analysis) theory is adopted in the integration of semi-finished ontology. The most obvious benefit of FCA is that the implicit link of. concept can be obtained without the influence of the different opinions of the participants. We develop the ontology integration system based on FCA. Firstly, we convert the ontology to formal context of one-value. Secondly, simplify the attributes of formal context using the method of discernibility matrix, in order to improve the combining efficiency of FCA. Moreover, redundant attributes of the formal concept are removed to improve the efficiency. Finally, the FCA of clear hierarchical structure is utilized to generate a unified ontology target.