Concept Algebra And Its Application In Ontology Merging

The Semantic Web, also known as the "next generation Internet", is an extension of the current Internet, its goal is to achieve enabling people to communicate with machines through machine understanding networking information. Ontology is the main support of the Semantic Web, which is used to describe information in semantic and knowledge levels, and to achieve the goal of machine understanding the information. Because different ontology development team use different development language and have different focuses on describing information, so that ontology heterogeneity which dooms to reduce knowledge reusability is resulted in. To resolve the problem of ontology heterogeneity, ontology merging as an effective way is chosen as an essential alternative solution in this thesis.However, ontology still has some shortcomings, such as concept expression depth insufficiency, low formalized degree, and can not facilitate machines to mimic the artificial ontology building process. In order to compensate for the ontology defects mentioned above and make improvement on knowledge expression, concept algebra is introduced, which proposed to represent knowledge based on cognitive theory. Concept Algebra has a higher formalization degree, and can reflect consciousness activity when experts building ontology, and easily achieves machine-imitating human thinking. Combining concept algebra with ontology, a novel method of concept algebra based ontology merging is proposed.Firstly, ontology is represented as concept network through concept algebra. Secondly, to reduce the numbers of following concept merging or associated operation and improve ontology merging efficiency, a method of semantic-associated degree based concept network region partition is proposed to divide concept network into several regions. Thirdly, the inter-region matching between two region-divided concept networks is carried out, region-matched degree is calculated, and a method of concept algebra based concept network fusion is proposed. Fourthly, using region-matched degree as a criterion, the inter-region concepts between two regions with a biggest region-matched degree should be merged or associated with the operators of concept algebra, so that inter-concept-network fusions can be realized. Fifthly, the fused concept networks are transformed into corresponding OWL-formed ontologies. Finally, a concept algebra based ontology merging system named as CABOMS is designed and implemented.To examine the performance of the proposed model and verify the validity of the designed prototype system, an experiment is conducted by using ontologies provided by OAEI2012. The experimental results reveal that the performance of the proposed model is reliable and the proposed CABOMS system has not only high merging-accurateness rate but also high merging-completeness rate. Therefore, the proposed method in this thesis provides a novel approach to ontology merging.