Research map / Graph learning / GNN architectures
Retrieval-Augmented Generation for Code Summarization via Hybrid GNN
Shangqing Liu, Yu Chen, Xiaofei Xie, Jing Kai Siow, Yang Liu
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use retrieval-augmented hybrid graph generation rather than Rencos | CCSD overall code summarization · CCSD overall | BLEU-4: 12.59 -> 14.01 | Improved |
| use retrieval-augmented hybrid graph generation rather than Rencos | CCSD overall code summarization · CCSD overall | ROUGE-L: 28.45 -> 30.89 | Improved |
| use retrieval-augmented hybrid graph generation rather than Rencos | CCSD overall code summarization · CCSD overall | METEOR: 13.21 -> 14.50 | Improved |
| add retrieval augmentation (ablation) | HGNN · CCSD overall | BLEU-4: 10.26 -> 14.01 | Improved |
| add retrieval augmentation (ablation) | HGNN · CCSD overall | ROUGE-L: 27.17 -> 30.89 | Improved |
| add static graph (ablation) | HGNN · CCSD overall | ROUGE-L: 30.47 -> 30.89 | Improved |
| add dynamic graph (ablation) | HGNN · CCSD overall | BLEU-4: 13.31 -> 14.01 | Improved |
| add both static and dynamic graphs (ablation) | HGNN without retrieval augmentation · CCSD overall | BLEU-4: 9.98 -> 10.26; 9.93 -> 10.26 | Improved |
| use summary-based rather than code-based retrieval augmentation (ablation) | HGNN augmentation · CCSD overall | BLEU-4: 10.22 -> 13.76 | Improved |
| combine code-based and summary-based retrieval augmentation (ablation) | HGNN augmentation · CCSD overall | BLEU-4: 14.01 | Improved |
| use HGNN rather than Rencos | PCSD code summarization · PCSD test | ROUGE-L: 36.21 -> 39.91 | Improved |
| remove static graph (ablation) | HGNN · PCSD test | BLEU-4: 24.42 -> 24.06 | Hurt |
Closest papers by meaning
- PinNet: Pinpoint Instructive Information for Retrieval Augmented Code-to-Text Generation (ICML 2024 · similarity 0.86)
- Towards A Generalist Code Embedding Model Based On Massive Data Synthesis (NeurIPS 2025 · similarity 0.84)
- CoRNStack: High-Quality Contrastive Data for Better Code Retrieval and Reranking (ICLR 2025 · similarity 0.83)
- CONSIDER: Commonalities and Specialties Driven Multilingual Code Retrieval Framework (AAAI 2024 · similarity 0.83)
- A Systematic Exploration of Knowledge Graph Alignment with Large Language Models in Retrieval Augmented Generation (AAAI 2025 · similarity 0.83)
- GFM-RAG: Graph Foundation Model for Retrieval Augmented Generation (NeurIPS 2025 · similarity 0.83)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
