2020/10/18 by Maliheh Izadi, Abbas Heydarnoori, Izadi, Maliheh +3 · 2 citations
Computer Science · Social Sciences · #FOS: Computer and information sciences #Software Engineering (cs.SE) #Software Engineering Research #Web Data Mining and Analysis #Wikis in Education and Collaboration
paper · pdf · doi:10.48550/arxiv.2010.09116
openalex publication_date 2020/10/18 · openalex created_date 2022/07/25 · openalex updated_date 2026/07/28
Many platforms exploit collaborative tagging to provide their users with\nfaster and more accurate results while searching or navigating. Tags can\ncommunicate different concepts such as the main features, technologies,\nfunctionality, and the goal of a software repository. Recently, GitHub has\nenabled users to annotate repositories with topic tags. It has also provided a\nset of featured topics, and their possible aliases carefully curated with the\nhelp of the community. This creates the opportunity to use this initial seed of\ntopics to automatically annotate all remaining repositories, by training models\nthat recommend high-quality topic tags to developers.\n In this work, we study the application of multi-label classification\ntechniques to predict software repositories' topics. First, we map the large\nspace of user-defined topics to those featured by GitHub. The core idea is to\nderive more information from projects' available documentation. Our data\ncontains about 152K GitHub repositories and 228 featured topics. Then, we\napply supervised models on repositories' textual information such as\ndescriptions, README files, wiki pages, and file names. We assess the\nperformance of our approach both quantitatively and qualitatively. Our proposed\nmodel achieves Recall@5 and LRAP scores of 0.890 and 0.805, respectively.\nMoreover, based on users' assessment, our approach is highly capable of\nrecommending a correct and complete set of topics. Finally, we use our models\nto develop an online tool named \Repository Catalogue, that\nautomatically predicts topics for GitHub repositories and is publicly\navailable.\n