The Research On Ontology-based Knowledge Representation

Knowledge representation is the key problem of knowledge engineering and the closest part in combination with artificial intelligence(AI). In traditional knowledge representation method, it is hard for knowledge to share because knowledge resource lacks united semantic description. Users can not make full use of knowledge they really need, so it is difficult for related sources to realize semantic fusion. How to solve these problems is the challenge knowledge engineering faces.According to the above disadvantages existed in traditional knowledge representation, this paper relying on teaching domain takes a research to knowledge representation based on ontology. First of all, after analyzing the shortcomings of traditional knowledge representation, the paper probes into the ontological knowledge representation, investigates the constructing method of knowledge ontology in teaching domain, and builds knowledge ontology of computer network (CNCO). The concepts in domain and relations among these concepts, such as inheritance relationship, instance relationship and restrictions, are built by the ontology to represent highly sharable and reusable domain knowledge and to meet the demand of knowledge representation in teaching domain.To realize knowledge share is mainly to solve the problem of ontology's heterogeneity. Ontology's heterogeneity arises because construction of ontology has no unified standard, which causes production of lots of different ontologies in the same domain. This paper has a deep research on ontology mapping which can solve the problem of ontology heterogeneity. And the key of ontology mapping is the computation of similarity. In this paper, it has an investigation from two aspects: conceptually semantic similarity and semantic relativity. Through combining the two aspects to have a measurement, the paper puts forward an evaluating similarity algorithm. Verification shows that this improved algorithm better solves the problem of measurement and computation between qualitative conception's similarity and relevance in traditional knowledge representation, and improves the accuracy and precision of similarity's evaluation computation. Based on this point, this paper further studies creation of similar matrix, establishment and carrying out of mapping rules and foundation of knowledge share's modeling frame, which lays a foundation for the final building of knowledge ontology mapping and for the realization of knowledge share.