Recommending posts concerning API issues in developer q&a sites

摘要: API design is known to be a challenging craft, as designers must balance their elegant ideals against "real-world" concerns, such utility, performance, backwards compatibility, and unforeseen emergent uses. However, date, there no principled method collect or analyze usability information that incorporates input from typical developers. In practice, developers often turn QaA websites stackoverflow.com (SO) when seeking expert advice on use, the popularity of sites has thus led very large volume unstructured can searched with diligence for answers specific questions. The collected wisdom within could, in principle, great help better support developer needs, if only it could collected, analyzed, distilled practical use. this paper, we present methodology combines several techniques, including social network analysis topic mining, recommend SO posts are likely concern design-related issues. To establish comparison baseline, introduce two more recommendation approaches: reputation-based recommender random recommender. We have found applied discussion popular mobile platforms, Android iOS, our achieves up 93% accuracy stable its recommendations compared baseline techniques.