CatRAG

KO | EN

gitlite — search

#javascript #python #hacktoberfest #typescript #llm #rust #macos #java #awesome #vue #pytorch #ios #openai #open-source

(★ 27)

CatRAG is a RAG framework builds on the HippoRAG 2 architecture and transforms the static KG into query-adaptive navigation structure. RAG + Knowledge Graphs + Context-Aware + Personalized PageRank. Accepted to Findings of ACL 2026.

CatRAG 최신버젼 다운로드

최종 버전 다운로드 (.zip)

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

<h1 align="center"> CatRAG: Context-Aware Transversal for Robust RAG </h1>

### CatRAG is a RAG framework builds on the HippoRAG 2 architecture and transforms the static KG into query-adaptive navigation structure.

This repository provides the **prompts** and **Hover dataset** for the paper *Breaking the Static Graph: Context-Aware Traversal for Robust Retrieval-Augmented Generation*.

Code will be released in a later update.

<p align="center">
  <img align="center" src="https://github.com/kwunhang/CatRAG/raw/main/images/catrag_method.png" />
</p>
<p align="center">
  <b>Comparison of graph traversal between HippoRAG 2 and CatRAG.</b> We illustrate the retrieval process for the multi-hop query “Which university did Marie Curie’s doctoral advisor attend?”. In HippoRAG 2 (top), the static graph structure causes semantic drift; probability mass is diverted to high-weight generic edges (e.g., *Marie Curie* → *Radioactivity*), missing the downstream evidence *ENS*. CatRAG (bottom) prevents this by applying (1) Symbolic Anchoring, injecting “University” as a weak seed, (2) Query-Aware Dynamic Edge Weighting, amplifying relevant paths (e.g., *Attend in ENS*) while pruning irrelevant ones, and (3) Key-Fact Passage Weight Enhancement, boosting relevant context edge strength. This steers the random walk to successfully retrieve the complete evidence chain for *ENS*.
</p>

## Latest News
* [2026-04-14] Accepted to Findings of ACL 2026.

### Check out our papers to learn more:
* [**Breaking the Static Graph: Context-Aware Traversal for Robust Retrieval-Augmented Generation**](https://arxiv.org/abs/2602.01965)

## What is included

- Prompts used in the paper experiments (see `prompts/`).
- Dataset / evaluation instances used in the paper (see `reproduce/`).

## Repository structure

- `prompts/`  
  Prompt templates and/or prompt sets used for evaluation.

- `reproduce/`  
  Evaluation dataset files (queries, documents/corpus references, and ground truth where applicable).

- `images/`  
  Images used in this repository.

- `README.md`  
  This document.

## Citation

If you find this work useful, please consider citing our papers:
```
@misc{lau2026breakingstaticgraphcontextaware,
      title={Breaking the Static Graph: Context-Aware Traversal for Robust Retrieval-Augmented Generation}, 
      author={Kwun Hang Lau and Fangyuan Zhang and Boyu Ruan and Yingli Zhou and Qintian Guo and Ruiyuan Zhang and Xiaofang Zhou},
      year={2026},
      eprint={2602.01965},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2602.01965}, 
}
```
## TODO:

- [ ] Add CatRAG logic code

Please feel free to open an issue or PR if you have any questions or suggestions.

CatRAG

CatRAG 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

CatRAG

CatRAG 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

Download README