Research map / Large language models / Retrieval and knowledge
DTKG: Dual-Track Knowledge Graph-Verified Reasoning Framework for Multi-Hop QA
Changhao Wang, Yanfang Liu, Xinxin Fan, Ao Tian, Lanzhi Zhou, Yunfeng Lu
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 few-shot question classifier routing to dual-track branches versus only fact verification (ablation) | DTKG · CWQ | ACC: 86.0 -> 90.0 | Improved |
| add classifier routing versus only chained reasoning branch (ablation) | DTKG · Mintaka | EM: 57.0 -> 67.6 | Improved |
| use accurate classifier instead of random branch assignment (ablation) | DTKG · HotpotQA | ACC: 73.0 -> 85.8 | Improved |
| add reranker stage to cosine-similarity scoring (two-stage scoring) (ablation) | DTKG · HotpotQA | ACC: 77.5 -> 85.8 | Improved |
| add cosine prefilter to reranker-only scoring (ablation) | DTKG · HotpotQA | ACC: 80.2 -> 85.8 | Improved |
| add two-stage scoring versus random relation selection (ablation) | DTKG · HotpotQA | ACC: 58.2 -> 85.8 | Improved |
| add generic rule-based denoising (ablation) | DTKG · CWQ | ACC: 75.8 -> 84.6 | Improved |
| use task-aware LLM denoising instead of generic denoising (ablation) | DTKG · CWQ | ACC: 84.6 -> 90.0 | Improved |
| add task-aware denoising on QALD10-en (ablation) | DTKG · QALD10-en | EM: 49.8 -> 50.0 | No clear effect |
| use DTKG instead of ToG path exploration on parallel-heavy dataset | Llama 3 8B · Mintaka | EM: 58.5 -> 67.6 | Improved |
| use DTKG instead of CoT on GraphRAG-Bench | Llama 3 8B · GraphRAG-Bench | EM: 14.4 -> 14.5 | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- CRITIC — CRITIC: Large Language Models Can Self-Correct with Tool-Interactive Critiquing (ICLR 2024) · better on 1 of 1 shared result
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
- Tree-of-Reasoning Question Decomposition for Complex Question Answering with Large Language Models (AAAI 2024 · similarity 0.90)
- Knowledge Editing with Dynamic Knowledge Graphs for Multi-Hop Question Answering (AAAI 2025 · similarity 0.89)
- Diagnosing and Addressing Pitfalls in KG-RAG Datasets: Toward More Reliable Benchmarking (NeurIPS 2025 · similarity 0.89)
- CofCA: A STEP-WISE Counterfactual Multi-hop QA benchmark (ICLR 2025 · similarity 0.89)
- Harnessing Large Language Models for Knowledge Graph Question Answering via Adaptive Multi-Aspect Retrieval-Augmentation (AAAI 2025 · similarity 0.88)
- Question Calibration and Multi-Hop Modeling for Temporal Question Answering (AAAI 2024 · similarity 0.88)
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.
