(A) clustering based schema matching scheme for improving matching correctness

Abstract

Nowadays, many functionality similar services are provided through heterogeneous interfaces because they are developed by different vendors and companies, and this makes an interface incompatibility problem unless users have devices which support multiple interfaces. Service adapter is a typical approach for solving problem happened by heterogeneous interfaces. It enables interoperability among heterogeneous interfaces by resolving incompatibility between them. However, as the number of services increases, the number of necessary adapters increases steeply. Therefore, developing adapters require too much burden of the adapter developers. For reducing effort of the adapter developer, there exist many approaches about adapter generation. For applying those approaches, interface matching should be done first. Also, schema matching is the one of widely used scheme for interface matching. However, schema matching has limitation because it uses only interface signature of the services. It cannot match interface well, if they are described using different words, so their signature is different. For solving this problem, we propose a refined schema matching scheme based on clustering of the methods. This approach reveals implicit information about relation between method and parameter. By comparing the method clusters, we can know the similarity of the parameters, and this leads improvement on the matching correctness and detecting the false matching.