Research On Framework And Approach Of Semantic Interoperation In Distributed Knowledge Management

The 21st century is a century of knowledge economy, and knowledge management has attracted more attentions from the enterprises. Meanwhile, with the increasingly emergences of large mounts of multinational companies and cooperation organizations (such as virtual organization, knowledge alliances and learning organization), distributed knowledge management (DKM) has gradually become an ad hoc topic. There are many evidences showing that the distributed characteristic of DKM is similar to the Semantic Web and what the Semantic Web focuses on is seen as an important solution for curing the heterogeneous problem of DKM. Taking these factors into consideration, this paper first presents a Framework for distributed knowledge management towards semantic interoperation (SiDKM), which mainly intends to solve the heterogeneous problem of DKM. Then the ontology mapping approach, the most important module in SiDKM, is seen as an key research object.The core work in this paper mainly involves the following threefold:(1) A Framework of distributed knowledge management towards semantic interoperation (SiDKM) is proposed. According to the existing heterogeneous problems of DKM, ontology theory in the Semantic Web is introduced. Specially, the three branches of ontology theory, namely, ontology building, ontology mapping and ontology evolution are all imported into the Framework of knowledge management. The aforementioned three branches are used for semantically constructing the knowledge base, sharing the knowledge and updating the knowledge respectively. In essence, this Framework extends the traditional knowledge management with the characteristic of language level processing by means of string processing into the new knowledge management style with the characteristic of semantic level processing by means of managing ontology. The new knowledge management style is more accessible to the principles in which human understands and reconstruct the subjects and can further improve the performances of knowledge management for enterprises.(2) The ontology mapping approach, the critical module in SiDKM, is deeply explored. By means of adopting ontologies to represent the enterprises' (organizations') knowledge bases, the existing heterogeneous problem of knowledge bases is transformed into that among ontologies. Ontology mapping appears to solve it. In this paper, from the systematic view, an integrated semi-automatic ontology mapping approach is presented based on three dimensions of linguistics, structures and instances combining commonsense knowledge with domain knowledge. The above three dimensions are integrated by means of weight approach and sigmoid function.(3) A distributed semantic retrieval prototype system is designed and implemented. Based on SiDKM and the proposed ontology mapping approach-IAOM, this prototype introduces the existing semantic similarity approach based on geometry distance to realize the valid semantic retrieval for heterogeneous distributed resources.In the end, several experiments are designed to test the precision and recall of IAOM and the performances of the proposed distributed semantic retrieval prototype system. The final results show the validity of the IAOM and the feasibility of SiDKM.