Only diff Is Not Enough: Generating Commit Messages Leveraging Reasoning and Action of Large Language Model
Problem
Automatically generating high-quality, informative commit messages for code changes, where prior approaches produced insufficient, poorly readable output.
Why it matters
Poor commit messages hinder software development and maintenance.
What the AI did
An LLM prompted with ReAct reasoning retrieves broad software context to generate factually grounded commit messages, incorporating quality expectations gathered from practitioners.
Results
Human evaluation demonstrates overall effectiveness; no concrete numbers stated in abstract
Impact
Human evaluation indicates the approach improves commit message quality over prior techniques.
Record
| Venue | Proceedings of the ACM on software engineering |
|---|---|
| Published | 2024 |
| Work type | article |
| Authors | Jiawei Li, David Faragó, Christian Petrov, Iftekhar Ahmed |
| Citations | 21 (OpenAlex, at import time) |
| References | 57 |
| Model or method | ReAct-prompted large language model; specific model not stated in source |
| Organization | not stated in source |
| Primary source | https://doi.org/10.1145/3643760 |
| Open access | hybrid |
| Keywords | commit message generation, large language models, ReAct prompting, software context, software engineering |
Importance rating
incremental / niche
Commit messages matter for software development and maintenance, but this is a narrow developer-tooling task; human evaluation shows improvement, no scale of affected users stated.
More in Software & Coding
| # | Paper | Year | Citations | Importance |
|---|---|---|---|---|
| 1 | ImageNet classification with deep convolutional neural networks | 2017 | 106,345 |
8
|
| 2 | Deep Residual Learning for Image Recognition | 2016 | 221,134 |
8
|
| 3 | Faster sorting algorithms discovered using deep reinforcement learning | 2023 | 175 |
8
|
| 4 | A deep learning‐based approach for software vulnerability detection using code metrics | 2022 | 42 |
7
|
| 5 | Deep Learning-Based Attack Detection and Classification in Android Devices | 2023 | 41 |
7
|