A Neural Network Based Schema Matching Method for Web Service Matching

With the development of Web service technology, Web service matching is becoming more and more important to service composition and service Mashup. The research of service matching has made great development and the crucial problem is schema matching. In this paper, we consider the task of matching multiple Web services for discovering the similar function services to the given one. In order to get the accurate matching result, we propose a neural network based schema matching (NNSM) method. While the semantic similarity calculated by individual matchers relies on individual aspects of information about schemas only, which are not sufficient for finding element correspondences between schemas. Our method uses a feed-forward neural network to combine multiple matchers where the back propagation algorithm is used to train the network. This brings significant gains in coverage while yielding modest gains in relevance. The experimental result shows we have a better accuracy compared to individual matchers and related research on service matching task.