In Vigder, M. & Chechik, M., editors, Proceedings of the 18<sup>th</sup> IBM Centers for Advanced Studies Conference (CASCON), pages 23-37, October, 2008. ACM Press. Most Influential Paper Award at CASCON'18. 15 pages.Paper abstract bibtex
Bug tracking systems are valuable assets for managing maintenance activities. They are widely used in open-source projects as well as in the software industry. They collect many different kinds of issues: requests for defect fixing, enhancements, refactoring/restructuring activities and organizational issues. These different kinds of issues are simply labeled as ``bug" for lack of a better classification support or of knowledge about the possible kinds. This paper investigates whether the text of the issues posted in bug tracking systems is enough to classify them into corrective maintenance and other kinds of activities. We show that alternating decision trees, naive Bayes classifiers, and logistic regression can be used to accurately distinguish bugs from other kinds of issues. Results from empirical studies performed on issues for Mozilla, Eclipse, and JBoss indicate that issues can be classified with between 77% and 82% of correct decisions.