Do Automatic Comment Generation Techniques Fall Short? Exploring the Influence of Method Dependencies on Code Understanding

• URL: http://arxiv.org/abs/2504.19459v1
• Date: Mon, 28 Apr 2025 03:49:06 GMT
• Title: Do Automatic Comment Generation Techniques Fall Short? Exploring the Influence of Method Dependencies on Code Understanding
• Authors: Md Mustakim Billah, Md Shamimur Rahman, Banani Roy,
• Abstract summary: Method-level comments are critical for improving code comprehension and supporting software maintenance.<n>This study investigates the prevalence and impact of dependent methods in software projects and introduces a dependency-aware approach for method-level comment generation.<n>We propose HelpCOM, a novel dependency-aware technique that incorporates helper method information to improve comment clarity, comprehensiveness, and relevance.
• Score: 1.971759811837406
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Method-level comments are critical for improving code comprehension and supporting software maintenance. With advancements in large language models (LLMs), automated comment generation has become a major research focus. However, existing approaches often overlook method dependencies, where one method relies on or calls others, affecting comment quality and code understandability. This study investigates the prevalence and impact of dependent methods in software projects and introduces a dependency-aware approach for method-level comment generation. Analyzing a dataset of 10 popular Java GitHub projects, we found that dependent methods account for 69.25% of all methods and exhibit higher engagement and change proneness compared to independent methods. Across 448K dependent and 199K independent methods, we observed that state-of-the-art fine-tuned models (e.g., CodeT5+, CodeBERT) struggle to generate comprehensive comments for dependent methods, a trend also reflected in LLM-based approaches like ASAP. To address this, we propose HelpCOM, a novel dependency-aware technique that incorporates helper method information to improve comment clarity, comprehensiveness, and relevance. Experiments show that HelpCOM outperforms baseline methods by 5.6% to 50.4% across syntactic (e.g., BLEU), semantic (e.g., SentenceBERT), and LLM-based evaluation metrics. A survey of 156 software practitioners further confirms that HelpCOM significantly improves the comprehensibility of code involving dependent methods, highlighting its potential to enhance documentation, maintainability, and developer productivity in large-scale systems.

Related papers

• Static Program Analysis Guided LLM Based Unit Test Generation [2.977347176343005]
  We describe a novel approach to automating unit test generation for Java methods using large language models (LLMs)<n>We show that augmenting prompts with emphconcise and emphprecise context information obtained by program analysis %of the focal method increases the effectiveness of generating unit test code through LLMs.
  arXiv  Detail & Related papers  (2025-03-07T13:09:37Z)
• Enhancing Input-Label Mapping in In-Context Learning with Contrastive Decoding [71.01099784480597]
  Large language models (LLMs) excel at a range of tasks through in-context learning (ICL) We introduce In-Context Contrastive Decoding (ICCD), a novel method that emphasizes input-label mapping. ICCD emphasizes input-label mapping by contrasting the output distributions between positive and negative in-context examples.
  arXiv  Detail & Related papers  (2025-02-19T14:04:46Z)
• A Controlled Study on Long Context Extension and Generalization in LLMs [85.4758128256142]
  Broad textual understanding and in-context learning require language models that utilize full document contexts. Due to the implementation challenges associated with directly training long-context models, many methods have been proposed for extending models to handle long contexts. We implement a controlled protocol for extension methods with a standardized evaluation, utilizing consistent base models and extension data.
  arXiv  Detail & Related papers  (2024-09-18T17:53:17Z)
• Icing on the Cake: Automatic Code Summarization at Ericsson [4.145468342589189]
  We evaluate the performance of an approach called Automatic Semantic Augmentation of Prompts (ASAP) We compare the performance of four simpler approaches that do not require static program analysis, information retrieval, or the presence of exemplars.
  arXiv  Detail & Related papers  (2024-08-19T06:49:04Z)
• Evaluating Saliency Explanations in NLP by Crowdsourcing [25.763227978763908]
  We propose a new human-based method to evaluate saliency methods in NLP by crowdsourcing. We recruited 800 crowd workers and empirically evaluated seven saliency methods on two datasets with the proposed method. We analyzed the performance of saliency methods, compared our results with existing automated evaluation methods, and identified notable differences between NLP and computer vision (CV) fields when using saliency methods.
  arXiv  Detail & Related papers  (2024-05-17T13:27:45Z)
• Comments as Natural Logic Pivots: Improve Code Generation via Comment Perspective [85.48043537327258]
  We propose MANGO (comMents As Natural loGic pivOts), including a comment contrastive training strategy and a corresponding logical comment decoding strategy. Results indicate that MANGO significantly improves the code pass rate based on the strong baselines. The robustness of the logical comment decoding strategy is notably higher than the Chain-of-thoughts prompting.
  arXiv  Detail & Related papers  (2024-04-11T08:30:46Z)
• Masked Thought: Simply Masking Partial Reasoning Steps Can Improve Mathematical Reasoning Learning of Language Models [102.72940700598055]
  In reasoning tasks, even a minor error can cascade into inaccurate results. We develop a method that avoids introducing external resources, relying instead on perturbations to the input. Our training approach randomly masks certain tokens within the chain of thought, a technique we found to be particularly effective for reasoning tasks.
  arXiv  Detail & Related papers  (2024-03-04T16:21:54Z)
• Generating Java Methods: An Empirical Assessment of Four AI-Based Code Assistants [5.32539007352208]
  We assess the effectiveness of four popular AI-based code assistants, namely GitHub Copilot, Tabnine, ChatGPT, and Google Bard. Results show that Copilot is often more accurate than other techniques, yet none of the assistants is completely subsumed by the rest of the approaches.
  arXiv  Detail & Related papers  (2024-02-13T12:59:20Z)
• Kun: Answer Polishment for Chinese Self-Alignment with Instruction Back-Translation [30.053409671898933]
  Kun is a novel approach for creating high-quality instruction-tuning datasets for large language models (LLMs) without relying on manual annotations. We leverage unlabelled data from diverse sources such as Wudao, Wanjuan, and SkyPile to generate a substantial dataset of over a million Chinese instructional data points.
  arXiv  Detail & Related papers  (2024-01-12T09:56:57Z)
• Unity is Strength: Cross-Task Knowledge Distillation to Improve Code Review Generation [0.9208007322096533]
  We propose a novel deep-learning architecture, DISCOREV, based on cross-task knowledge distillation. In our approach, the fine-tuning of the comment generation model is guided by the code refinement model. Our results show that our approach generates better review comments as measured by the BLEU score.
  arXiv  Detail & Related papers  (2023-09-06T21:10:33Z)
• Self-Polish: Enhance Reasoning in Large Language Models via Problem Refinement [50.62461749446111]
  Self-Polish (SP) is a novel method that facilitates the model's reasoning by guiding it to progressively refine the given problems to be more comprehensible and solvable. SP is to all other prompting methods of answer/reasoning side like CoT, allowing for seamless integration with state-of-the-art techniques for further improvement.
  arXiv  Detail & Related papers  (2023-05-23T19:58:30Z)
• FewNLU: Benchmarking State-of-the-Art Methods for Few-Shot Natural Language Understanding [89.92513889132825]
  We introduce an evaluation framework that improves previous evaluation procedures in three key aspects, i.e., test performance, dev-test correlation, and stability. We open-source our toolkit, FewNLU, that implements our evaluation framework along with a number of state-of-the-art methods.
  arXiv  Detail & Related papers  (2021-09-27T00:57:30Z)
• Towards Improved and Interpretable Deep Metric Learning via Attentive Grouping [103.71992720794421]
  Grouping has been commonly used in deep metric learning for computing diverse features. We propose an improved and interpretable grouping method to be integrated flexibly with any metric learning framework.
  arXiv  Detail & Related papers  (2020-11-17T19:08:24Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.