Expert recommendation in oss projects based on knowledge embedding

Modern Open Source Software (OSS) projects depend on the globally-distributed and synchronized software development. The online collaboration promotes more and more developers to join in OSS projects, while on the other hand, integrating new developers with teams is challenging and pivotal to the success of a project. In this paper, we propose a novel expert recommendation method, based on knowledge embedding, that realizes real-time recommendation for working developers. To capture structural information of source files in call graph, we use node2vec algorithm to convert file entities within projects into knowledge mappings within low-dimensional space, based on which we further propose four features to capture the work status and social relationship of developers. We then design a recommender system using random forest method to recommend appropriate experts for the developers. Experiments on 20 Apache OSS projects show that, compared with the baseline methods, our approach behaves significantly better in terms of a series of performance metrics.