Deep Learning and Search-Based Approach for Duplicate Bug Report Detection in Industrial Datasets
Fabiano Tavares da Silva, Felipe Rocha de Araújo, Diego Falcão de Souza, Erick C. Bezerra · 2023
The large-scale production of software naturally produces releases, which in turn are susceptible to bugs. These bugs are then reported in bug tracking systems, leading to occurrence reports that reference the same underlying defect, known as duplicate bug reports. Identifying these duplicates poses a significant challenge, particularly in the context of global-scale software development, where a substantial volume of bug reports are reported. To address this problem, existing literature has explored the use of pairs of bugs, treating them as either duplicates or non-duplicates and leveraging classification techniques. However, when considering the practical implementation of such approaches in a production environment, the process of comparing a new bug report to the entire collection of prior bug reports becomes time-consuming and resource-intensive. Consequently, applying them in industrial environments becomes unfeasible. To overcome this challenge, this paper proposes a method that combines deep learning and search-based techniques, making the applicability of such method feasible for large datasets. It is based on a real industry case, and the results proved the approach to be useful in industrial environments.