CNN Based Severity Prediction of Bug Reports

Abstract

Users can use bug tracking systems (BTS) such as Bugzilla, JIRA, and GitHub to report defects they detect while using or testing the software applications. Bugzilla has specified the bug’s information such as bug ID, description, product, classification, component, platform, operating system, bug status, resolution, priority, and severity. The severity assigned to the bug report is conducted manually. So, it takes some considerable time to assign the severity manually, and also it is a tedious task. In this research, we used a deep learning-based technique to develop the prediction model for severity. Our proposed approach is based on the convolutional neural network (CNN) for severity prediction of the bug reports. First, we pre-process the textual content in bug reports using natural language processing (NLP) approaches. Then we extract the features from the textual context (short description) using the Bag-of-Words feature extraction method. Finally, we train a CNN-based classifier to severity prediction based on its input. Then our result is compared with the Support Vector Machine (SVM) and Temporal Convolutional Network (TCN) to find a better model for severity prediction. The final results show that the proposed approach based on the CNN classifier performs better than the other approaches, and it shows an 81% accuracy while others have low accuracy, like 61% and 48% for SVM and TCN, respectively. The proposed model’s performance was evaluated using the Bugzilla dataset, which included over 25,000 bug reports. The accuracy, precision, recall, and F1-Score are used to measure the performance.