Research On Concept Similarity Algorithm In OWL DL Ontology

As a model describing information of semantic web,ontology represents information semantic by the perfect layered structure of concept and class,and the attributed relationships between primitives and concepts,and support to logic reasoning.Since ontology provides an expressive form with semantic information, the information processing in the world of web is based on operation of ontology. Operations to different ontologies are based on the ontology similarity and partial similarity,so it is important to research on concept similarity in ontologies.Through analysis and summary of the current calculation of concept similarity in ontologies,it is found that T-H BACH algorithm gets satisfactory calculation results of similarity by using the primitives representing semantic information of concepts in OWL DL ontology.But T-H BACH algorithm is not perfect while calculating similarity,which lies in the following aspects,for example,it needs to be changed of the adjusting strategy of the weight in the semantic relativity of the primitive,the calculating semantic of similarity is lost because of ignorance of influence on similarity from hierarchical relationship and so on.Aiming at the above problems,this thesis proposes the SOC-MATCH algorithm,which is also a method calculating concept in ontology similarity. Basing on the T-H BACH algorithm that primitive of expression describing logic is taken as the element of calculation of concept similarity;the SOC-MATCH algorithm improves the weigh adjusting strategy of primitive,and comprehensively considering the influence of layered structure to concept in ontologies,calculates the similarity with regard to semantic distance,contact ration of concept and offspring hierarchy.Comparing SOC-MATCH with T-H BACH algorithm at evaluating indicator of discrepancy definition,it is found that SOC-MATCH algorithm retains the algorithm character of T-H BACH algorithm and in the meanwhile,SOC-MATCH algorithm increases calculating precision of concept similarity in ontology,having stronger ability differentiating concept similarity in the same level and its precision of discrepancy definition is better than T-H BACH algorithm.SOC-MATCH algorithm is able to calculate concept similarity in ontologies conveying the same semantic information quickly and accurately when calculating concept similarity in ontologies,reflecting semantic information more correct and comprehensive,having plentiful expression for concept similarity and the result satisfies people's sensation.According to SOC-MATCH algorithm,a comprehensive calculating model for concept similarity in ontologies is designed and realized,providing calculating functions such as semantic similarity,structural similarity of concept, T-H BACH similarity and SOC-MATCH similarity,it offers the computer aid to research on concept similarity in ontologies,and provides support for research areas such as ontology mapping,ontology integration and semantic searches and so on.