Students from our laboratory presented their research at ESEM2026.
Ms. Shirakawa and Ms. Yoshimoto from our laboratory presented the following two papers at The 20th ACM/IEEE International Symposium on Empirical Software Engineering and Measurement (ESEM 2026), held in Munich, Germany, in 2026.
ESEM is an international conference dedicated to empirical research in software engineering and is considered a top-tier conference; this year marks its 20th edition. For the technical paper track—which saw a 75% increase in submissions compared to the previous year—51 papers were accepted out of 269 submissions (an acceptance rate of 18.9%).
Ayane Shirakawa, Tatsuya Shirai, Yutaro Kashiwa, Masanari Kondo, Yasutaka Kamei, Hajimu Iida, “Test Alert Snooze: An Empirical Study of Consecutive Test Failures on CI”, In Proceedings of the 20th ACM/IEEE International Symposium on Empirical Software Engineering and Measurement (ESEM 2026)
Ms. Shirakawa presented the findings of this study, which aimed to clarify the realities of consecutive test failures in Continuous Integration (CI) and the associated developer behaviors. The study defined a “Test Alert Snooze” — the state in which the same test method fails across two or more consecutive revisions on CI — and empirically analyzed its occurrence rate, duration, and developer behavior. The results showed that Test Alert Snooze occurred in 4 out of the 27 OSS projects studied (approximately 15%). A detailed analysis of these four projects revealed that 61.1% of the instances involved failures across two consecutive revisions, while 52.9% were resolved within one day. Furthermore, an analysis of developer behavior based on the classification of revisions using CSS types indicated a tendency for documentation to be modified while a Test Alert Snooze was active, whereas tests were more likely to be modified when the issue was resolved. (LIPIcs.ESEM.2026.24)
Suzuka Yoshimoto, Kosei Horikawa, Daniel Feitosa, Yutaro Kashiwa, Hajimu Iida, “Is Self-Admitted Technical Debt Tested? An Empirical Study of Coverage, Co-change, and Impact”, In Proceedings of the 20th ACM/IEEE International Symposium on Empirical Software Engineering and Measurement (ESEM 2026)
Ms. Yoshimoto presented the findings of this study, which investigated the relationship between Self-Admitted Technical Debt (SATD)—technical debt explicitly acknowledged by developers through comments such as TODO or FIXME—and testing. The study analyzed eight Java projects, examining test coverage for 784 SATD instances in the latest releases, as well as test additions and bug-induction rates associated with 5,175 SATD resolutions. The results showed that existing tests covered 60.7% of SATD-containing code, while only 3.4% of commits resolving SATD added new tests targeting the modified code. There was no substantial difference in bug-induction rates 30–180 days after SATD resolution depending on whether tests were added; over a longer period, however, the rate was higher when tests were added (6.32% vs. 4.37%). These findings suggest that developers tend to treat SATD resolution as an ordinary code change rather than as a high-risk maintenance activity. (LIPIcs.ESEM.2026.36)
