Predicting Merge Conflicts Considering Social and Technical Assets

This technical report investigates the prediction of merge conflicts in software development by considering both social and technical assets. The study identifies the challenges developers face when resolving merge conflicts, which are often time-consuming and error-prone. The authors propose three models to predict these conflicts based on various developer roles and their contributions. The report details a large-scale empirical study involving 66 software repositories and 78,740 merge scenarios, categorizing developers into top and occasional contributors. Findings indicate that top contributors at the project level and occasional contributors at the merge-scenario level are more likely to cause conflicts. The report emphasizes the importance of understanding developer roles in minimizing merge conflicts and presents machine learning techniques that achieve high accuracy in predictions. The results suggest that incorporating social measures can enhance the prediction of merge conflicts, which is crucial for improving collaborative software development practices.