Automatic Commit Message Generation: A Critical Review and Directions for Future Work
Open accessThe DOI matches its Crossref record; 19 of the 19 references it cites were found in a registry or an index.
Abstract
Commit messages are critical for code comprehension and software maintenance. Writing a high-quality message requires skill and effort. To support developers and reduce their effort on this task, several approaches have been proposed to automatically generate commit messages. Despite the promising performance reported, we have identified three significant and prevalent threats in these automated approaches: 1) the datasets used to train and evaluate these approaches contain a considerable amount of ‘noise’; 2) current approaches only consider commits of a limited diff size; and 3) current approaches can only generate the subject of a commit message, not the message body. The first limitation may let the models ‘learn’ inappropriate messages in the training stage, and also lead to inflated performance results in their evaluation. The other two threats can considerably weaken the practical usability of these approaches. Further, with the rapid emergence of large language models (LLMs) that show superior performance in many software engineering tasks, it is worth asking: can LLMs address the challenge of long diffs and whole message generation? This article first reports the results of an empirical study to assess the impact of these three threats on the performance of the state-of-the-art auto generators of commit messages. We collected commit data of the Top 1,000 most-starred Java projects in GitHub and systematically removed noisy commits with bot-submitted and meaningless messages. We then compared the performance of four approaches representative of the state-of-the-art before and after the removal of noisy messages, or with different lengths of commit diffs. We also conducted a qualitative survey with developers to investigate their perspectives on simply generating message subjects. Finally, we evaluate the performance of two representative LLMs, namely UniXcoder and ChatGPT, in generating more practical commit messages. The results demonstrate that generating commit messages is of great practical value, considerable work is needed to mature the current state-of-the-art, and LLMs can be an avenue worth trying to address the current limitations. Our analyses provide insights for future work to achieve better performance in practice.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Language Models are Few-Shot Learners
Brown, T. B., Mann, B., Ryder, N., Subbiah, M., Kaplan, J., Dhariwal, P., Neelakantan, A., Shyam, P., Sastry, G., Askell, A., et al. (2020) · arXiv · doi:10.48550/arxiv.2005.14165 · Cited by 23,372
Effect size estimates: Current use, calculations, and interpretation.
Fritz, C. O., Morris, P. E., & Richler, J. J. (2012) · Journal of Experimental Psychology: General · doi:10.1037/a0024338 · Cited by 5,199 · cited 1×
“If a statistically significant difference ( p .05) was found, we also calculated the effect size, which provides a measure of how large the difference is [65].”
BackgroundA Survey on Evaluation of Large Language Models
Chang, Y., Wang, X., Wang, J., Wu, Y., Yang, L., Zhu, K., Chen, H., Yi, X., Wang, C., Wang, Y., et al. (2024) · ACM Transactions on Intelligent Systems and Technology · doi:10.1145/3641289 · Cited by 4,254 · cited 1×
“Large language models (LLMs) have garnered significant attention and adoption in both academic and industrial domains [42], including Software Engineering (SE) [17], [43], [44], due to their exceptional performance across a wide range of applications.”
BackgroundSelf-citationCodeBERT: A Pre-Trained Model for Programming and Natural Languages
Feng, Z., Guo, D., Tang, D., Duan, N., Feng, X., Gong, M., Shou, L., Qin, B., Liu, T., Jiang, D., et al. (2020) · Findings of the Association for Computational Linguistics: EMNLP 2020 · doi:10.18653/v1/2020.findings-emnlp.139 · Cited by 2,824
PaLM: Scaling Language Modeling with Pathways
Chowdhery, A., Narang, S., Devlin, J., Bosma, M., Mishra, G., Roberts, A., Barham, P., Chung, H. W., Sutton, C., Gehrmann, S., et al. (2022) · arXiv · doi:10.48550/arxiv.2204.02311 · Cited by 2,590
Wang, Y., Wang, W., Joty, S., & Hoi, S. C. H. (2021) · Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing · doi:10.18653/v1/2021.emnlp-main.685 · Cited by 1,495 · cited 1×
“There are also some pre-trained models designed for encoding diff, such as [18], [26], which are usually evaluated in the task of commit message generation.”
BackgroundWang, Y., Wang, W., Joty, S., & Hoi, S. C. H. (2021) · arXiv · doi:10.48550/arxiv.2109.00859 · Cited by 1,495
A Prompt Pattern Catalog to Enhance Prompt Engineering with ChatGPT
White, J., Fu, Q., Hays, S., Sandborn, M., Olea, C., Gilbert, H., Elnashar, A., Spencer-Smith, J., & Schmidt, D. C. (2023) · arXiv · doi:10.48550/arxiv.2302.11382 · Cited by 838
Lin, C.-Y., & Och, F. J. (2004) · Proceedings of the 42nd Annual Meeting on Association for Computational Linguistics - ACL '04 · doi:10.3115/1218955.1219032 · Cited by 737 · cited 1×
“2) ROUGE-L measures the F-score of precision and recall on the basis of the longest common subsequence (LCS) between a generated message and a reference message [57].”
BackgroundUniXcoder: Unified Cross-Modal Pre-training for Code Representation
Guo, D., Lu, S., Duan, N., Wang, Y., Zhou, M., & Yin, J. (2022) · Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · doi:10.18653/v1/2022.acl-long.499 · Cited by 673
Cited by
Do Advanced Language Models Eliminate the Need for Prompt Engineering in Software Engineering?
Wang, G., Sun, Z., Ye, S., Gong, Z., Chen, Y., Zhao, Y., Liang, Q., & Hao, D. (2026) · ACM Transactions on Software Engineering and Methodology · doi:10.1145/3771933 · Cited by 26
Metadata
- Authors
- Zhang, Y., Qiu, Z., Stol, K.-J., Zhu, W., Zhu, J., Tian, Y., & Liu, H.
- Venue
- IEEE Transactions on Software Engineering
- Published
- 2024-04
- Volume
- 50
- Issue
- 4
- Pages
- 816-835
- ISSN
- 0098-5589 (ISSN-L)
- 1939-3520
- 2326-3881
- OpenAlex
- W4391741735
- Cited by
- 42
- References
- 19
- FWCI
- 16.07
- Type
- journal article
- Language
- en
- Open access
- hybrid
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: Automatic Commit Message Generation: A Critical Review and Directions for Future Work. https://app.cochat.ai/papers/10.1109/tse.2024.3364675
- The paper
- Zhang, Y., Qiu, Z., Stol, K.-J., Zhu, W., Zhu, J., Tian, Y., & Liu, H. (2024). Automatic Commit Message Generation: A Critical Review and Directions for Future Work. IEEE Transactions on Software Engineering. https://doi.org/10.1109/tse.2024.3364675
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.