Mashup-Oriented API Recommendation via Random Walk on Knowledge Graph

With the growing prosperity of the Web API economy, mashup-oriented API recommendation has become an important requirement. Various methods based on different principles of technology have been used to deal with this issue. In recent years, the Web API ecosystem has accumulated a wealth of knowledge that can be used to enhance the recommendation models, and however, current concerns in this regard still remain. To cope with this issue, we present a graph-based algorithmic framework for the task of mashup-oriented API recommendation. Especially, we design a concise schema of the knowledge graph to encode the mashup-specific contexts and model the mashup requirement with graphic entities. We then exploit random walks with restart to assess the potential relevance between the mashup requirement and the Web APIs according to the knowledge graph. In addition, we propose the query-specific weighting strategies to enhance the knowledge graph construction. The experimental results demonstrate that our proposed method is much superior to some state-of-the-art methods, also achieves robust effects on reducing computational overhead, and suppresses the negative Matthew effect in APIs’ recommendation.