A Knowledge Graph Embedding Based Service Recommendation Method for Service-Based System Development

Web API is an efficient way for Service-based Software (SBS) development, and mashup a key technology which merges several web services to deal with the increasing complexity of software requirements expedite service-based system development. The service recommendation method vital However, existing methods often suffer from data sparsity or cold start issues, should lead bad effects. Currently, this paper starts SBS proposes based on knowledge graph embedding collaborative filtering (CF) technology. In our model, we first construct refined using SBS-service co-invocation record related information mine potential semantics relationship between service. Then, learn entities in graph. These heterogeneous (SBS service, etc.) are embedded into low-dimensional space through representation learning algorithms Word2vec TransR, distances vectors calculated. input model requirement (target SBS), similarities functional set extracted graph, can relieve problem. Meanwhile, uses CF recommend target SBS. Finally, verifies effectiveness real-word dataset. Compared state-of-the-art methods, has best hit rate ranking quality.