Service recommendation based on quotient space granularity analysis and covering algorithm on Spark

The rapid growth of Web services has made it a challenge for users to find appropriate Web services because it is very difficult for traditional Web service recommendation approaches to process the large amount of service-relevant data. To address this issue, this paper proposes CA-QGS (Covering Algorithm based on Quotient space Granularity analysis on Spark), a scalable approach for accurate Web service recommendation in large-scale scenarios. CA-QGS first clusters users and Web services based on users' past quality experiences on co-invoked Web services. It then performs granularity analysis on the clustering results to identify users and Web services that are similar to the target user and Web service, and employs the collaborate filtering technique to predict the quality of the target Web service for the target user. This way, appropriate Web services can finally be recommended to the target user. To increase the efficiency of CA-QGS, we parallelize CA-QGS on Spark. Extensive experiments show that CA-QGS outperforms existing approaches in both recommendation accuracy and efficiency.