ILinker: a novel approach for issue knowledge acquisition in GitHub projects

Abstract

Social coding facilitates the sharing of knowledge in GitHub projects. In particular, issue reports, as an important knowledge in the software development, usually contain relevant information, and can thus be shared and linked in the developers’ discussion to aid the issue resolution. Linking issues to potentially related issues, i.e. issue knowledge acquisition, would provide developers with more targeted resource and information when they search and resolve issues. However, identifying and acquiring related issues is in general challenging, because the real-world acquiring practice is time-consuming and mainly depends on the experience and knowledge of the individual developers. Therefore, acquiring related issues automatically is a meaningful task which can improve development efficiency of GitHub projects. In this paper, we formulate the problem of acquiring related issue knowledge as a recommendation problem. To solve this problem, we propose a novel approach, iLinker, combining information retrieval technique, i.e. TF-IDF, and deep learning techniques, i.e. Word Embedding and Document Embedding. Our evaluation results show that, in both coarse-grained recommendation and fine-grained recommendation tasks, iLinker outperforms the baseline approaches.