An Approach For Ontology Mapping Using Wordnet

The appearance of the World Wide Web has changed the way of information storage, expression, transmission and obtain. At present, the content of web page in www are mostly redacted in HTML. However, this web pages are suitable to read and handle for human beings while it is very difficult for machine to automatic data interoperation. In this situation, the concept of Semantic Web has been proposed. The tentative idea of Semantci Web is that the information are stored which can be dealt by computer, so that we can achive the goal of more intelligence information retrive and so on.With the more and more study about Semantic Web, the concept of ontology has been proposed. An ontology can give a formal description about a domain using concepts, relations, instances and so on. Because of the creator of an ontology is different, the composition of the ontology is heterogeneity. In order to achieve the aim of the semantic interoperation between the heterogeneity ontologies, it is necessary to implement ontology mapping so that we can find the semantic correspondence.The main part of this paper propose an approach for calculating semantic similarity between words using WordNet. This approach simulated the thought process of human when they judge two word is similar or not. The survey found that people prefer to consider more differences when the semantic distance between two words is closer, and vice versa. We use the semantic distances between two words in WordNet to signify the same and different points between two words.This paper use three different strategy to calculate the semantic similarity between two elements in different ontologies which including name similarity, structure similarity. Specially, we adopt the new approach for calculating similarity to compute the name similarity.In order to confirm the effectiveness of the new approach of calculating similarity, this paper extracted28noun pairs form D&G dataset as our test set. We also compute the correlation with the human judgment. The correlation between our experiment result and human rating is0.884. By the comparison with other existing method, we see that our approach outperforms others. In the same way, we adop the benchmark tests which is supported by the international organization OAEI. This benchmark tests include60test ontologies and a reference ontology. The test ontologies can be divided in several categories which are mostly obtained by discarding some information from the reference ontology. When it comes to evaluative criteria, we use the precision, recall and f-measure which are commonly using in the information retrive domain. The result shows that our mapping approach performas very well.In the process of ontology mapping study, we not only took example by the present literature, but also discover some existing problem. There are some problems in the area of ontology mapping still needed to slove. At the end of this paper, we indicate the future of the research. In the math area,there are not only1-1mappings, but also1:n mappings and so on. But most of the present literature just took account of the1-1mappings, and most of them were designed for English ontology. When it came to Chinese ontology, the study was little. Above of all, there is much room for improvement.