A social network approach for recommending interoperable Web services

Modern application development leverages the invocation of a large pool of Web services such as Cloud services and APIs. As the number of Web services keeps growing, it becomes difficult for developers to identify services that can collaborate as part of the same composite application, or that can replace each other in failure cases. Gathering and analyzing Web services social interaction such as composition, substitution, and subscription helps building communities of interoperable services (i.e., likely to collaborate with each other and/or to replace each other). This paper proposes a new approach for recommending interoperable services to developers based on the multi-dimensional analysis of their social interaction history. The approach aims to build communities of services with highly dense interaction relationships. Services part of the same community are recommended to developers as potential collaborators or substitutes. The proposed approach identifies first service leaders. Leaders are particular services with a high interaction rate in the network around which communities are built. Remaining services followers join communities based on their previous interaction experiences. Followers leverage the votes of their experienced neighbors to make their final vote. Experiments on pseudo-real data show that leveraging services social interaction outperforms state-of-the-art approaches.