Research map / Large language models / Retrieval and knowledge
Reason-Align-Respond: Aligning LLM Reasoning With Knowledge Graphs for KGQA
Xiangqing Shen, Fanfan Wang, Zinong Yang, Bing Wang, Wenli Du, Chengqing Zong, Rui Xia
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 LLM-driven Consolidation (ablation) | RAR · CWQ | F1: 83.0 -> 84.8 | Improved |
| add the Reasoner (reasoning chains) (ablation) | RAR · CWQ | F1: 80.3 -> 84.8 | Improved |
| add KG-constrained Decoding (ablation) | RAR · CWQ | F1: 48.9 -> 84.8 | Improved |
| add Knowledge Path Expansion (ablation) | RAR · CWQ | F1: 71.7 -> 84.8 | Improved |
| remove LLM-driven Consolidation: recall (ablation) | RAR · CWQ | recall: 89.0 -> 91.2 | Hurt |
| use Llama-3.1-8B instead of Llama-2-7B for Reasoner and Aligner (ablation) | RAR · CWQ | F1: 68.9 -> 72.6 | Improved |
| scale Reasoner/Aligner from Qwen2-0.5B to Qwen2-7B (ablation) | RAR · CWQ | Hit: 71.3 -> 81.4 | Improved |
| use GPT-4o instead of GPT-4o-mini as consolidator (ablation) | RAR · CWQ | Hit: 91.0 -> 92.8 | Improved |
| use Llama-3.1-70B instead of GPT-4o-mini as consolidator (ablation) | RAR · CWQ | F1: 84.8 -> 83.0 | Hurt |
| increase beam size (ablation) | RAR · CWQ | Hit/F1 | Improved |
| run more EM iterations (ablation) | RAR · CWQ | Hit/F1 | Improved |
| use RAR instead of GCR | RAR · CWQ | Hit: 75.8 -> 91.0 | Improved |
Closest papers by meaning
- Reasoning on Graphs: Faithful and Interpretable Large Language Model Reasoning (ICLR 2024 · similarity 0.90)
- Graph-constrained Reasoning: Faithful Reasoning on Knowledge Graphs with Large Language Models (ICML 2025 · similarity 0.89)
- Reasoning-as-Logic-Units: Scaling Test-Time Reasoning in Large Language Models Through Logic Unit Alignment (ICML 2025 · similarity 0.88)
- General-Reasoner: Advancing LLM Reasoning Across All Domains (NeurIPS 2025 · similarity 0.88)
- LightPROF: A Lightweight Reasoning Framework for Large Language Model on Knowledge Graph (AAAI 2025 · similarity 0.88)
- EPERM: An Evidence Path Enhanced Reasoning Model for Knowledge Graph Question and Answering (AAAI 2025 · similarity 0.87)
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.
