A concept analysis approach for guiding users in service discovery

Web services are widely used as basic constructs to build complex distributed applications with fast speed and low cost. However, existing service discovery techniques provide users with poor results which require substantial human intervention to filter the services to locate the desired ones. In particular, users often have no prior knowledge of the functional description of the available services on the Web. The queries formulated by the users may not match well with the service descriptions of existing services. As a consequence, a user's query can result in a large number of returned services. In this paper, we propose an approach that derives the semantic concepts conveyed in the service descriptions and clusters the services based on the concepts. As a result, each concept is associated with a set of relevant services. To understand the semantic meanings of a user's query, we identify concepts behind the query and recommend related concepts associated with services. Our approach also guides users to formulate their queries. We conducted a case study and found that the average precision and recall of our approach for service discovery are respectively, 83% and 100%. We also performed a user study which shows that for 85% of time, a user reformulates their queries using the suggestion provided by our approach to improve the precision of the retrieved services.