Title :
TRAM: An approach for assigning bug reports using their Metadata
Author :
Banitaan, Shadi ; Alenezi, M.
Author_Institution :
Dept. of Math., Comput. Sci. & Software Eng., Univ. of Detroit Mercy, Detroit, MI, USA
Abstract :
Bug triage is an essential phase in the bug fixing process. The aim of bug triage is to assign an experienced developer to a new coming bug report. Existing bug triage approaches are mainly based on machine learning techniques. These approaches suffer from low prediction accuracy. In this paper, we propose TRAM (TRiaging Approach using bug reports Metadata). The goal is to improve the prediction accuracy of bug triage by utilizing the most discriminating terms of bug reports, the components in which the bugs belong to, and the reporter who filed the bug.We perform experimental evaluation on open-source projects namely Freedesktop, NetBeans, Eclipse, and Firefox. The results show that TRAM outperforms existing machine learning-based approaches in terms of classification accuracy. TRAM improves the F-score by approximately 34%, 40%, 20%, and 21% for Freedesktop, NetBeans, Eclipse, and Firfox respectively.
Keywords :
learning (artificial intelligence); meta data; program debugging; Eclipse; Firefox; Firfox; Freedesktop; NetBeans; TRAM; bug fixing process; bug triage; machine learning; open-source projects; triaging approach using bug reports metadata; Accuracy; Computer bugs; Predictive models; Software; Software engineering; Training; Vectors; bug triage; classification; mining bug repositories; term selection method;
Conference_Titel :
Communications and Information Technology (ICCIT), 2013 Third International Conference on
Conference_Location :
Beirut
Print_ISBN :
978-1-4673-5306-9
DOI :
10.1109/ICCITechnology.2013.6579552
