Cross-Domain Developer Recommendation Algorithm Based on Feature Matching

In recent years, the software crowdsourcing has become a new software development pattern. More and more developers choose to publish, search for software tasks, and solve software problems on software crowdsourcing platform. As such, the platform generates a large amount of developer and development task information every day, which makes it difficult for developers to find appropriate tasks from massive tasks. Therefore, it is significant to deploy developer recommendation system on crowdsourcing platforms. Now, most developer recommendation algorithms can only use single platform data. Since the new software crowdsourcing platforms do not have enough historical behavior information of developers, previous developer recommendation algorithms cannot recommend developers to new tasks effectively. To solve the sparsity problem, this paper proposes a cross-domain developer recommendation algorithm based on feature matching. Firstly, we seek from the auxiliary domain for the most similar tasks to the current target domain task. Then, we retrieved the corresponding developers of these tasks. Finally, we select from the target domain the most similar developer to the developers retrieved to compose the recommendation developer set of the current task. In order to verify the effectiveness of the proposed algorithm, we crawls data from two different software crowdsourcing platforms to conduct experiments and compare the proposed model with various advanced developer recommendation algorithms. The experimental results show that the proposed algorithm has advantages over the previous algorithms on different evaluation metrics.