Similarity Analysis of Service Descriptions for Efficient Web Service Discovery

Web services are currently one of the preferred ways for realizing Service Oriented Architectures in business systems. Due to this popularity and also due partly to the failure of the Universal Business Registry initiative, the number of published service descriptions openly available on the Web has increased by a large extent and hence, relevant service discovery as per user specification remains a challenge. In order to achieve more efficient discovery, we propose a crawler based system for gathering service descriptions available on the Web for building a scalable service repository. We apply similarity analysis techniques to the service descriptions after extracting features provided by the service descriptions and automatically generate relevant tags for each service. Using Agglomerative Hierarchical clustering, we cluster the tagged service descriptions and use the same tagging technique to generate tags for each cluster. For generating cluster tags, we take into account how well the tag represents the corresponding service in the cluster and how well the service itself represents the cluster it is in. The search domain for service discovery was significantly reduced by tagging & clustering and and we show that our system achieves good results.