Automatic bug assignment has been a topic of interest for researchers in recent years. Textual bug reports play a crucial role in helping engineers fix bugs, but the presence of noise in these reports can negatively affect automatic bug assignment. Classical Natural Language Processing (NLP) techniques have limitations in handling textual features, leading to the need for improved techniques.
Zexuan Li and his research team delved into the effects of textual and nominal features on bug assignment approaches. Their study, published in the Frontiers of Computer Science, aimed to determine whether advanced NLP techniques can enhance the performance of textual features in bug assignments. The team employed TextCNN to analyze textual features and compared them with nominal features to assess their effectiveness.
Surprisingly, the results of the study indicated that textual features did not outperform nominal features, even with the use of advanced NLP techniques. The research team identified nominal features as influential in bug assignment approaches, as they indicated developers’ preferences. By conducting experiments with different classifiers and feature groups, they demonstrated that nominal features alone could achieve competitive results without relying on textual content.
The research study focused on three main questions:
1. The effectiveness of textual features when combined with deep-learning-based NLP techniques.
2. The identification of influential features for bug assignment approaches and the reasons behind their significance.
3. The extent to which selected influential features can enhance bug assignments.
Despite the utilization of improved NLP techniques like TextCNN, the study found that the impact on bug assignment performance was limited. The selected key nominal features showed an accuracy improvement ranging from 11% to 25% under popular classifiers such as Decision Tree and SVM. The researchers suggested that future work could explore the integration of source files to establish a knowledge graph linking influential features and descriptive words for better embedding of nominal features. This approach could potentially enhance bug assignment accuracy and efficiency.
Overall, the research study sheds light on the importance of considering both textual and nominal features in automatic bug assignment processes. By understanding the influence of these features and their implications for bug assignment approaches, researchers and engineers can improve the effectiveness of bug-fixing procedures in software development.
Leave a Reply