TicketTrace: Intelligent High Parity Ticket Detection Through Deep Learning Techniques
Vijay Raj P, Jyoti Shetty · 2023
Bug reports play a crucial role in software development by facilitating communication between users and developers and aiding in the resolution of software defects. However, the presence of duplicate bug reports can lead to inefficiencies in the bug tracking process, resulting in wasted resources and delayed issue resolution. This paper propose a novel approach for detecting high parity bug reports using deep learning techniques. This paper's approach leverages the power of deep neural networks to automatically identify similarities and patterns within bug report data, the bug report that have used to train is taken from multiple software projects. Model is trained on a carefully curated dataset of bug reports, consisting of textual information such as titles and descriptions. The heart of this papers solution lies in the architecture of Siamese deep learning model. The trained siamese network is then used to take new inputs tickets and compare them with existing tickets to detect high parity ones. All the implementation uses open source python packages such as tensorflow, keras and transformers to build the neural network. In the evaluation phase, the performance of proposed model is measured using established metrics, including precision, recall and F1-score. The empirical findings illustrate that the utilization of deep learning techniques led to an accuracy of 79 and 81 percent for two distinct testsets. Additionally elaborate on the constraints of the paper's methodology and discuss potential enhancements for the model implemented.