Reproduce “EvoRAG: Making Knowledge Graph-based RAG Automatically Evolve through Feedback Backpropagation” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Zhenbo Fu, Yuanzhe Zhang, Qiange Wang, Hao Yuan, Yuehao Xu, Enze Yi, Yanfeng Zhang, Ge Yu
EvoRAG adapts knowledge-graph retrieval-augmented generation by using response feedback to adjust the priorities of facts in a static knowledge graph. It evaluates reasoning paths, then propagates their utility to triplets, changing their likelihood of future retrieval. Across three datasets, the authors report an average 7.34% accuracy improvement over KG-RAG frameworks; the method does not verify or correct underlying facts.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.