Leveraging Track Relationships for Web Service Recommendation

Existing Web services recommendation approaches are based on usage statistics or QoS properties, leaving aside the evolution of the services' ecosystem. These approaches do not always capture new or more recent users' preferences resulting in recommendations with possibly obsolete or less relevant services. In this paper, we describe a novel Web services recommendation approach where the services' ecosystem is represented as a heterogeneous multi-graph, and edges may have different semantics. The recommendation process relies on data mining techniques to suggest services "of interest" to a user.