Code Review Automation using Retrieval Augmented Generation

2025/11/07 by Meng, Qianru, Zhang, Xiao, Ren, Zhaochen +1
#FOS: Computer and information sciences #Software Engineering (cs.SE)

paper · doi:10.48550/arxiv.2511.05302

Abstract

Code review is essential for maintaining software quality but is labor-intensive. Automated code review generation offers a promising solution to this challenge. Both deep learning-based generative techniques and retrieval-based methods have demonstrated strong performance in this task. However, despite these advancements, there are still some limitations where generated reviews can be either off-point or overly general. To address these issues, we introduce Retrieval-Augmented Reviewer (RARe), which leverages Retrieval-Augmented Generation (RAG) to combine retrieval-based and generative methods, explicitly incorporating external domain knowledge into the code review process. RARe uses a dense retriever to select the most relevant reviews from the codebase, which then enrich the input for a neural generator, utilizing the contextual learning capacity of large language models (LLMs), to produce the final review. RARe outperforms state-of-the-art methods on two benchmark datasets, achieving BLEU-4 scores of 12.32 and 12.96, respectively. Its effectiveness is further validated through a detailed human evaluation and a case study using an interpretability tool, demonstrating its practical utility and reliability.

Citations

• LAURA: Enhancing Code Review Generation with Context-Enriched Retrieval-Augmented LLM
• SARA: Selective and Adaptive Retrieval-augmented Generation with Context Compression
• Retrieval-Augmented Code Review Comment Generation
• PISCO: Pretty Simple Compression for Retrieval-Augmented Generation
• EXIT: Context-Aware Extractive Compression for Enhancing Retrieval-Augmented Generation
• Retrieval-Augmented Semantic Parsing: Improving Generalization with Lexical Knowledge
• RAG-Enhanced Commit Message Generation
• Retrieval-Augmented Generation for Large Language Models: A Survey
• LLaMA-Reviewer: Advancing Code Review Automation with Large Language Models through Parameter-Efficient Fine-Tuning
• Diverse Retrieval-Augmented In-Context Learning for Dialogue State Tracking
• Unified Demonstration Retriever for In-Context Learning
• Lift Yourself Up: Retrieval-augmented Text Generation with Self Memory
• Suggestion Bot: Analyzing the Impact of Automated Suggested Changes on Code Reviews
• Large Language Models are Few-Shot Summarizers: Multi-Intent Comment Generation via In-Context Learning
• AUGER: Automatically Generating Review Comments with Pre-training Models
• Automating Code Review Activities by Large-Scale Pre-training
• ReACC: A Retrieval-Augmented Code Completion Framework
• Using Pre-Trained Models to Boost Code Review Automation
• CodeT5: Identifier-aware Unified Pre-trained Encoder-Decoder Models for Code Understanding and Generation
• Retrieval Augmented Code Generation and Summarization
• Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks
• Augmenting Transformers with KNN-Based Composite Memory for Dialogue
• Template-Based Question Generation from Retrieved Sentences for Improved Unsupervised Question Answering
• Dense Passage Retrieval for Open-Domain Question Answering
• REALM: Retrieval-Augmented Language Model Pre-Training
• CORE: Automating Review Recommendation for Code Changes
• Attention Is All You Need

Related

• Combining Large Language Models with Static Analyzers for Code Review Generation
• LAURA: Enhancing Code Review Generation with Context-Enriched Retrieval-Augmented LLM
• Retrieval-Augmented Code Review Comment Generation
• Deep Learning-based Code Reviews: A Paradigm Shift or a Double-Edged Sword?
• Deep Assessment of Code Review Generation Approaches: Beyond Lexical Similarity