Context-Based Web Service Clustering

Under web service environment, services´ calling or called relationships are usually represented as links. Based on link- or link-content analysis, it may produce tight clusters, which is useful for resource management. However current work only considers the existence of collaboration and it seldom takes service usage context into consideration. In this paper, we suppose to cluster similar services by introducing service usage context which is the complete stories for services´ activities and helps to improve clustering performance. Being different with link- or link-content-based clustering methods that considering one-linkdistance-away-neighbors (part of the activity story), this work proves that in team-up work situation, all collaborators (long-distance-neighbor) can provide information for distinguishing services.