Research map / Large language models / Retrieval and knowledge
Stepwise Contrastive Reasoning for Retrieval-Augmented Generation over Knowledge Graphs
Chenxiao Lin, Ye Luo, KunHong Liu, Qingqiang Wu
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 |
|---|---|---|---|
| add stepwise reasoning (with multi-step retrieval) (ablation) | SCR · WebQSP | F1: 45.4 -> 74.8 | Improved |
| add stepwise reasoning (with multi-step retrieval) (ablation) | SCR · WebQSP | Hit: 70.1 -> 90.8 | Improved |
| add stepwise reasoning (with multi-step retrieval) (ablation) | SCR · CWQ | F1: 30.5 -> 60.2 | Improved |
| add stepwise reasoning (with multi-step retrieval) (ablation) | SCR · CWQ | Hit: 56.5 -> 70.4 | Improved |
| add beam search (ablation) | SCR · WebQSP | F1: 49.1 -> 74.8 | Improved |
| add beam search (ablation) | SCR · CWQ | F1: 38.2 -> 60.2 | Improved |
| use customized relation-aware GNN instead of R-GCN (ablation) | SCR · WebQSP | F1: 66.8 -> 74.8 | Improved |
| use customized relation-aware GNN instead of R-GCN (ablation) | SCR · CWQ | F1: 57.0 -> 60.2 | Improved |
| increase beam size K from 1 to larger values (ablation) | SCR · WebQSP | Hit | Improved |
| increase beam size K beyond 5 (ablation) | SCR · WebQSP | F1 | Hurt |
| use SCR instead of GNN-RAG on multi-hop CWQ questions | SCR · CWQ >=3-hop | F1: 51.8 -> 64.2 | Improved |
| use SCR instead of SubgraphRAG | SCR · WebQSP | F1: 77.5 -> 74.8 | Hurt |
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)
- A Systematic Exploration of Knowledge Graph Alignment with Large Language Models in Retrieval Augmented Generation (AAAI 2025 · similarity 0.92)
- Think-on-Graph 2.0: Deep and Faithful Large Language Model Reasoning with Knowledge-guided Retrieval Augmented Generation (ICLR 2025 · similarity 0.92)
- Harnessing Large Language Models for Knowledge Graph Question Answering via Adaptive Multi-Aspect Retrieval-Augmentation (AAAI 2025 · similarity 0.91)
- Can Knowledge-Graph-based Retrieval Augmented Generation Really Retrieve What You Need? (NeurIPS 2025 · similarity 0.90)
- Retrieval is Not Enough: Enhancing RAG through Test-Time Critique and Optimization (NeurIPS 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.
