SOFTWARE DATA REDUCTION FOR EFFECTIVE BUG TRIAGE

Computer science

Software companies experience a significant decline of over 45% in revenue due to software issues. Bug triage plays a vital role in the bug-fixing process by assigning developers to new problems. To streamline and reduce manual effort in bug triage, automatic bug triage uses text categorization algorithms. Our research focuses on data reduction for bug triage, aiming to decrease data size while enhancing quality. To achieve this, we combine instance and feature selection techniques to reduce the size of bug reports and individual words. Using variables from existing bug datasets, we develop a prediction model to determine the optimal sequence for applying instance and feature selection on a new bug dataset. To validate our approach, we conduct an empirical analysis on a dataset comprising 600,000 bug reports from Eclipse and Mozilla, two well-known open-source projects. Our study's findings demonstrate that our proposed data reduction methodology effectively decreases data size while improving bug triage accuracy. This research offers a valuable technique for employing data processing methods to generate concise yet high-quality bug data in software development and maintenance

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.