Socially-enriched semantic mashup of web APIs

As Web mashups are becoming one of the salient tools for providing composite services that satisfy users' requests, there have been many endeavors to enhance the process of recommending the most adequate mashup to users. However, previous approaches show numerous pitfalls such as the problem of cold-start, and the lack of utilization of social information as well as functional properties of Web APIs and mashups. All these factors undoubtedly hinder the proliferation of mashup users as locating the most appropriate mashup becomes a cumbersome task. In order to resolve the issues, we propose an efficient method of recommending mashups based on the functional and social features of Web APIs. Specifically, the proposed method utilizes the social and functional relationships among Web APIs to produce and recommend the chains of candidate mashups. Experimental results with a real world data set show a precision of 86.9% and a recall of 75.2% on average, which validates that the proposed method performs more efficiently for various kinds of user requests as compared to a previous work.