TY - GEN
T1 - DeepJIT
T2 - 16th IEEE/ACM International Conference on Mining Software Repositories, MSR 2019
AU - Hoang, Thong
AU - Khanh Dam, Hoa
AU - Kamei, Yasutaka
AU - Lo, David
AU - Ubayashi, Naoyasu
N1 - Funding Information:
Our future work involves extending our evaluation to other open source and commercial projects. We also plan to extend DeepJIT using attention neural network [78] so that our model can explain its predictions to software practitioners. We also plan to implement DeepJIT into a tool (e.g. a GitHub plugin) to assess its usefulness in practice. Dataset and Code. The dataset and code for DeepJIT are available at https://github.com/AnonymousAccountConf/ DeepJTT MSR. Acknowledgements. This research was partially supported by Singapore National Research Foundation (award number: NRF2016-NRF-ANR003) and JSPS KAKENHI Grant Numbers JP15H05306 and JP18H0322.
Publisher Copyright:
© 2019 IEEE.
PY - 2019/5
Y1 - 2019/5
N2 - Software quality assurance efforts often focus on identifying defective code. To find likely defective code early, change-level defect prediction - aka. Just-In-Time (JIT) defect prediction - has been proposed. JIT defect prediction models identify likely defective changes and they are trained using machine learning techniques with the assumption that historical changes are similar to future ones. Most existing JIT defect prediction approaches make use of manually engineered features. Unlike those approaches, in this paper, we propose an end-to-end deep learning framework, named DeepJIT, that automatically extracts features from commit messages and code changes and use them to identify defects. Experiments on two popular software projects (i.e., QT and OPENSTACK) on three evaluation settings (i.e., cross-validation, short-period, and long-period) show that the best variant of DeepJIT (DeepJIT-Combined), compared with the best performing state-of-the-art approach, achieves improvements of 10.36-11.02% for the project QT and 9.51-13.69% for the project OPENSTACK in terms of the Area Under the Curve (AUC).
AB - Software quality assurance efforts often focus on identifying defective code. To find likely defective code early, change-level defect prediction - aka. Just-In-Time (JIT) defect prediction - has been proposed. JIT defect prediction models identify likely defective changes and they are trained using machine learning techniques with the assumption that historical changes are similar to future ones. Most existing JIT defect prediction approaches make use of manually engineered features. Unlike those approaches, in this paper, we propose an end-to-end deep learning framework, named DeepJIT, that automatically extracts features from commit messages and code changes and use them to identify defects. Experiments on two popular software projects (i.e., QT and OPENSTACK) on three evaluation settings (i.e., cross-validation, short-period, and long-period) show that the best variant of DeepJIT (DeepJIT-Combined), compared with the best performing state-of-the-art approach, achieves improvements of 10.36-11.02% for the project QT and 9.51-13.69% for the project OPENSTACK in terms of the Area Under the Curve (AUC).
UR - http://www.scopus.com/inward/record.url?scp=85072337768&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85072337768&partnerID=8YFLogxK
U2 - 10.1109/MSR.2019.00016
DO - 10.1109/MSR.2019.00016
M3 - Conference contribution
AN - SCOPUS:85072337768
T3 - IEEE International Working Conference on Mining Software Repositories
SP - 34
EP - 45
BT - Proceedings - 2019 IEEE/ACM 16th International Conference on Mining Software Repositories, MSR 2019
PB - IEEE Computer Society
Y2 - 26 May 2019 through 27 May 2019
ER -