Research On Ontology Matching Key Technology Based On WordNet

Ontology technology has been used extensively as the core of Semantic Web. However, the development of ontology is distributed, that is, different users will develop different ontologies even for the same domain. Thus, there are many different ontologies in a certain domain. These ontologies are heterogeneous, which can't communicate with each other. In order to use and reuse ontology to complete the task of information exchange and integration more effectively, we must find the mapping between different ontologies, which is ontology matching technique.Ontology Matching is the key technology to address the communication between heterogeneous ontologies and find out the entity elements mapping relationship of different ontologies. It greatly contributes to achieve the communication between the applications based on different ontologies. As the incensement of ontology application requirement, the number of ontology is increasing sharply, so the methods and tools to complete ontology matching automatically or semi-automatic are needed urgently.In this paper, we research on ontology matching. And we have respectively studied the ontology matching algorithms from the element level and structure level. Main achievements are described as follows:Firstly, this paper researches the similarity algorithm which is the key technique of ontology matching. After analyzing the strongpoint and short coming of existing similarity algorithm, we proposed a similarity measure algorithm based on both WordNet and direct graph theory. In order to validate algorithm's performance, we used the classical data to verify the similarity algorithm. The results show that our algorithm has obvious advantages.Secondly, according to the structural characteristics of the ontology attributes and relations, we respectively proposed a similarity calculation method based on ontology matching.Finally, we design a composite framework of ontology matching based on similarity calculation. It computes the similarities of both element level and structure level, then give the final similarity of the matching pair with the weighted average. We use OAEI's experimental cases to validate our framework, analyzing the recall rate, precision rate and F-Measure. The experiment results show that the proposed matching algorithm has a better matching quality and performance.