e-journal
An Effective Web Service Ranking Method via Exploring User Behavior
Abstract
Service-oriented computing and Web services are becoming more and more popular, enabling organizations to use the Web as a market for selling their own Web services and consuming existing Web services from others. Nevertheless, with the increasing adoption and presence of Web services, it becomes more difficult to find the most appropriate Web service that satisfies both users’ functional and nonfunctional requirements. In this paper, we propose an effective Web service ranking approach based on collaborative filtering (CF) by exploring the user behavior, in which the invocation and query history are used to infer the potential user behavior. CF-based user similarity is calculated
through similar invocations and similar queries (including functional query and QoS query) between users. Three aspects of Web services—functional relevance, CF based score, and QoS utility, are all considered for the final Web service ranking. To avoid the impact of different units, range, and distribution of variables, three ranks are calculated for the three factors respectively. The final Web service ranking is obtained by using a rank aggregation method based on rank positions. We also propose
effective evaluation metrics to evaluate our approach. Large-scale experiments are conducted based on a real world Web service dataset. Experimental results show that the proposed approach outperforms the existing approach on the rank performance.
Index Terms—Web service, service ranking, functional relevance, collaborative filtering, QoS utility, user behavior
Tidak ada salinan data
Tidak tersedia versi lain