Research Of Methods For Similarity Among Domain Ontology Concept Based On Concept Lattice

Ontology is widely used in many related fields as the expression of the shared concept knowledge model. As for domain ontology, there are inevitably such defection that the same concept has different definitions, resulting in the occurrence of ontology heterogeneous. The problem makes it hard to share and reuse the domain knowledge among ontologies, which severely restricted the scope of application domain ontology. Therefore, it is very essential to study ontology heterogeneous, and reduces the difficulty of interaction between ontologies.To achieve the knowledge sharing of domain ontologies, the common bridge should be built. Concept is the core area of knowledge representation as the important element of ontology, carrying huge amounts of information. So the establishment of the relationship among ontology concepts is the key to achieve the fusion and coupling of domain knowledge, large-scale collaborative domain ontology.This paper focuses on the similarity computation of concepts in the field of agriculture ontology. Especially three key problems in the extracting model of conceptual similarity are prospected, including resolving heterogeneity of domain ontology, standardizing formal context and extracting the similarity of concepts. In order to solve these problems above, theories and methods of extracting the concept similarity for the agricultural domain are proposed and corresponding prototype is also developed.The main research of this content is as followings:â‘ The digestion strategy theory of resolving heterogeneity among domain ontology is analyzed. Firstly, we investigates both the ontology description language and the method to uniformly standardize domain ontology representation. Secondly, the strategy set is built through resolving the concepts and properties of ontology. As a result, it is achieved to adjust the heterogeneous characteristics of the domain ontology language layer.â‘¡The method of nomalizing field forms is proposed. By transforming the set of ontology concept set and property set into the object-property hierarchical relationships, the domain formal context related to the similar area is obtained. Therefore, multiple relationships can be translated into binary relations. By set conversion, the combined formal background set is obtained.â‘¢The theory and method of calculating conceptual similarity is presented. The missing-valued relationship in formal context is mainly prospected, through dealing with the complete information of the background knowledge, the extraction algorithm based on the concept rank priorities of attribute weight is given, accessing to the set of concepts with a collection of different levels. Finally, the optimal similarity calculation model with the relation matrix is proposed.â‘£Based on the theory and method above, the concept similarity extracting prototype is developed under the programming environment of visual C++6.0, which is applied into the tea field. Experimental results show that the model improves computational accuracy, and reduces the numbers of irrelevant concepts. So the method and theory proposed is verified.