Research map / Large language models / Retrieval and knowledge
You Don’t Need Pre-Built Graphs for RAG: Retrieval Augmented Generation with Adaptive Reasoning Structures
Shengyuan Chen, Chuang Zhou, Zheng Yuan, Qinggang Zhang, Zeyang Cui, Hao Chen, Yilin Xiao, Jiannong Cao, Xiao Huang
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 |
|---|---|---|---|
| replace pre-built graph RAG with query-time logic DAG decomposition | LogicRAG vs HippoRAG2 · 2WikiMQA | string accuracy: 50.0 -> 64.7 | Improved |
| replace pre-built graph RAG with query-time logic DAG decomposition | LogicRAG vs HippoRAG2 · MuSiQue | string accuracy: 27.0 -> 30.4 | Improved |
| replace pre-built graph RAG with query-time logic DAG decomposition | LogicRAG vs HippoRAG2 · HotpotQA | string accuracy: 56.7 -> 54.8 string; 61.9 -> 62.6 LLM | Mixed |
| add retrieval over zero-shot GPT-4o-mini (vanilla RAG top-5) | GPT-4o-mini · HotpotQA | string accuracy: 38.7 -> 44.1 | Improved |
| increase top-k in vanilla RAG (ablation) | VanillaRAG · HotpotQA, 2WikiMQA, MuSiQue | string accuracy: HotpotQA 38.4 -> 44.1 (top-1 to top-5) | Improved |
| increase top-k beyond 5 (top-10 to top-20) (ablation) | LogicRAG · 2WikiMQA | accuracy | No clear effect |
| sampling without replacement of subproblem batches instead of with replacement (ablation) | LogicRAG · HotpotQA, 2WikiMQA, MuSiQue | token cost per question | Improved |
| add graph pruning (merge same-rank subproblems into one query) (ablation) | LogicRAG · multi-hop QA | retrieval rounds | Improved |
| add context pruning via rolling memory summary (ablation) | LogicRAG · multi-hop QA | token cost | Improved |
| use LogicRAG instead of graph-based RAG at query time | GraphRAG/LightRAG/HippoRAG2 · 2WikiMQA | average tokens: HippoRAG2 2809.2 -> LogicRAG 1777.9 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- HippoRAG2 — HippoRAG: Neurobiologically Inspired Long-Term Memory for Large Language Models (NeurIPS 2024) · better on 7 of 9 shared results
- RAPTOR — RAPTOR: Recursive Abstractive Processing for Tree-Organized Retrieval (ICLR 2024) · better on 2 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Simple is Effective: The Roles of Graphs and Large Language Models in Knowledge-Graph-Based Retrieval-Augmented Generation (ICLR 2025 · similarity 0.94)
- Think-on-Graph 2.0: Deep and Faithful Large Language Model Reasoning with Knowledge-guided Retrieval Augmented Generation (ICLR 2025 · similarity 0.93)
- StructRAG: Boosting Knowledge Intensive Reasoning of LLMs via Inference-time Hybrid Information Structurization (ICLR 2025 · similarity 0.91)
- From Complex to Atomic: Enhancing Augmented Generation via Knowledge-Aware Dual Rewriting and Reasoning (ICML 2025 · similarity 0.91)
- Retrieval is Not Enough: Enhancing RAG through Test-Time Critique and Optimization (NeurIPS 2025 · similarity 0.91)
- A Systematic Exploration of Knowledge Graph Alignment with Large Language Models in Retrieval Augmented Generation (AAAI 2025 · similarity 0.90)
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.
