A Quality-Aware Web API Recommender System for Mashup Development

The rapid increase in the number and diversity of web APIs with similar functionality, makes it challenging to find suitable ones for mashup development. In order to reduce the number of similarly functional web APIs, recommender systems are used. Various web API recommendation methods exist which attempt to improve recommendation accuracy, by mainly using some discovered relationships between web APIs and mashups. Such methods are basically incapable of recommending quality web APIs because they fail to incorporate web API quality in their recommender systems. In this work, we propose a method that considers the quality features of web APIs, to make quality web API recommendations. Our proposed method uses web API quality to estimate their relevance for recommendation. Specifically, we propose a matrix factorization method, with quality feature regularization, to make quality web API recommendations and also enhance recommendation diversity. We demonstrate the effectiveness of our method by conducting experiments on a real-world dataset from www.programmableweb.com. Our results not only show quality web API recommendations, but also, improved recommendation accuracy. In addition, our proposed method improves recommendation diversity by mitigating the negative Matthew effect of accumulated advantage, intrinsic to most existing web API recommender systems. We also compare our method with some baseline recommendation methods for validation.