Holistic Combination of Structural and Textual Code Information for Context Based API Recommendation

Context based API recommendation is an important way to help developers find the needed APIs effectively and efficiently. For effective recommendation, we need not only a joint view of both structural textual code information, but also holistic correlated usage in control data flow graph as whole. Unfortunately, existing methods exploit or information separately. In this work, propose novel approach called APIRec-CST (API Recommendation by Combining Structural Textual information). deep learning model that combines with text source on Graph Network Code Token simultaneously learn features for recommendation. We apply train JDK library 1,914 open-source Java projects evaluate accuracy MRR (Mean Reciprocal Rank) another 6 projects. The results show our achieves respectively top-1, top-5, top-10 60.3, 81.5, 87.7 69.4 percent, significantly outperforms graph-based statistical tree-based A further analysis shows makes sense improves MRR. sensitivity top-k are insensitive number be recommended hole. conduct user study which two groups students asked finish programming tasks without plugin. can faster more accurately feedback usability overwhelmingly positive.