Study On An Improved Semantic Matching Algorithm Based On Ontology

In information integration, semantic retrieval and ontology mapping among heterogenous information sources, it is a difficult problem to address semantic matching for a long time. Ontology can specify the concepts and relations among them. Therefore, an ontology-based method might be used to solve the semantic matching problem.The conflicts might emerge inevitably when there are more than one ontology in a system. How to resolve the conflicts is the goal of semantic matching.According to research, the matching between concepts from different ontologies can be discovered by their semantic similarity, And also in existing approaches, similarity often derives from instances, relations and information of hierarchy of concepts. After research we have discovered that the above three referenced objects reflect the relationship among concepts in some degree.Traditional semantic matching, the majority of which is based on keyword matching techniques, has inherent defects.This paper has involved some improvement against the existing approach, so that we got an integrated approach for computational model of semantic similarity based on instances, relations and hierarchy information of concepts.In this approach the similarity of concept can be defined by joint probability distribution of the concepts and instances. In the same way, the similarity of relation can be gotten by statistics and graph theory. The similarity of next concept can be found under the guide of the relation matching rule, then the similarity between concepts can be obtained by calling the function of concept similarity recursively.Finally all the values with weights of the concept similarity of every hierarchy should be added and then it can be integrated into the semantic similarity of two ontologies.In the paper, we built an ontology on fire apparatus domain, and finished an experiment about the semantic matching for the ontology. According to the result we compared precise degree of the original algorithms with the improved.The result indicates that the model can be used to calculate the semantic similarity of ontologies, and it makes some kind of improvement for efficiency of the recall and precision.In the experiment, the MPI class is imported into Java by using JNI technology,and we try to explore the approach of parallel computing in Java program environment.The experiment achieves the inter-operation between Java and C languages, and provides an effective solution for the model implement of parallel computing environment.