An Approach for Assigning Bug Reports Based on Developer Activities

Abstract:
In software project development, bug assignment is an important part of bug repair work. Manual defect assignment is inefficient. Now researchers have proposed many classification techniques to automatically assign bugs, but most of the research work pays little attention to the developer’s active level. This paper proposes a bug assignment method considering the developer’s active level. Our method uses the recurrent neural networks to extract the developer’s active level features, combine convolutional neural networks and recurrent neural networks to extract text features of bug reports, and merges the two features for bug assignment. Experimental results on Eclipse and Mozilla data sets show that our method has higher accuracy compared to other methods.