# Automatic Comment Generation via Multi-Pass Deliberation

arXiv: https://arxiv.org/abs/2209.06634 · HTML: https://mlsanity.com/paper/2209.06634/

## Key facts

- Cited by 3 papers in the map (data through 14 Oct 2026).
- Builds on 23 of its 65 references that are in the map.
- First posted on arXiv 14 Sep 2022.

## Builds on (most cited first, 23 of 23)

- [Attention Is All You Need](https://mlsanity.com/paper/1706.03762/) (2017)
- [Adam: A Method for Stochastic Optimization](https://mlsanity.com/paper/1412.6980/) (2014)
- [Layer Normalization](https://mlsanity.com/paper/1607.06450/) (2016)
- [CIDEr: Consensus-based Image Description Evaluation](https://mlsanity.com/paper/1411.5726/) (2014)
- [Circle Loss: A Unified Perspective of Pair Similarity Optimization](https://mlsanity.com/paper/2002.10857/) (2020)
- [A Transformer-based Approach for Source Code Summarization](https://mlsanity.com/paper/2005.00653/) (2020)
- [Sequence-to-Sequence Learning as Beam-Search Optimization](https://mlsanity.com/paper/1606.02960/) (2016)
- [Achieving Human Parity on Automatic Chinese to English News Translation](https://mlsanity.com/paper/1803.05567/) (2018)
- [ReACC: A Retrieval-Augmented Code Completion Framework](https://mlsanity.com/paper/2203.07722/) (2022)
- [Improving Automatic Source Code Summarization via Deep Reinforcement Learning](https://mlsanity.com/paper/1811.07234/) (2018)
- [A parallel corpus of Python functions and documentation strings for automated code documentation and code generation](https://mlsanity.com/paper/1707.02275/) (2017)
- [Deliberation Model Based Two-Pass End-to-End Speech Recognition](https://mlsanity.com/paper/2003.07962/) (2020)
- [Retrieve and Refine: Exemplar-based Neural Comment Generation](https://mlsanity.com/paper/1910.10419/) (2019)
- [Code Structure Guided Transformer for Source Code Summarization](https://mlsanity.com/paper/2104.09340/) (2021)
- [Code to Comment "Translation": Data, Metrics, Baselining & Evaluation](https://mlsanity.com/paper/2010.01410/) (2020)
- [Transformer Based Deliberation for Two-Pass Speech Recognition](https://mlsanity.com/paper/2101.11577/) (2021)
- [Adversarial Robustness of Deep Code Comment Generation](https://mlsanity.com/paper/2108.00213/) (2021)
- [Let's Ask Again: Refine Network for Automatic Question Generation](https://mlsanity.com/paper/1909.05355/) (2019)
- [Ensemble Models for Neural Source Code Summarization of Subroutines](https://mlsanity.com/paper/2107.11423/) (2021)
- [Retrieve and Refine: Exemplar-based Neural Comment Generation](https://mlsanity.com/paper/2010.04459/) (2020)
- [ROSF: Leveraging Information Retrieval and Supervised Learning for Recommending Code Snippets](https://mlsanity.com/paper/1703.01446/) (2017)
- [GN-Transformer: Fusing Sequence and Graph Representation for Improved Code Summarization](https://mlsanity.com/paper/2111.08874/) (2021)
- [Cross Modification Attention Based Deliberation Model for Image Captioning](https://mlsanity.com/paper/2109.08411/) (2021)

## Built on by (most cited first, 3 of 3)

- [Deep Learning for Code Intelligence: Survey, Benchmark and Toolkit](https://mlsanity.com/paper/2401.00288/) (2023)
- [Improving Retrieval-Augmented Code Comment Generation by Retrieving for Generation](https://mlsanity.com/paper/2408.03623/) (2024)
- [AI-Mediated Code Comment Improvement](https://mlsanity.com/paper/2505.09021/) (2025)

## Abstract (arXiv)

Deliberation is a common and natural behavior in human daily life. For example, when writing papers or articles, we usually first write drafts, and then iteratively polish them until satisfied. In light of such a human cognitive process, we propose DECOM, which is a multi-pass deliberation framework for automatic comment generation. DECOM consists of multiple Deliberation Models and one Evaluation Model. Given a code snippet, we first extract keywords from the code and retrieve a similar code fragment from a pre-defined corpus. Then, we treat the comment of the retrieved code as the initial draft and input it with the code and keywords into DECOM to start the iterative deliberation process. At each deliberation, the deliberation model polishes the draft and generates a new comment. The evaluation model measures the quality of the newly generated comment to determine whether to end the iterative process or not. When the iterative process is terminated, the best-generated comment will be selected as the target comment. Our approach is evaluated on two real-world datasets in Java (87K) and Python (108K), and experiment results show that our approach outperforms the state-of-the-art baselines. A human evaluation study also confirms the comments generated by DECOM tend to be more readable, informative, and useful.