Research map / Large language models / Retrieval and knowledge
Seeing through the Conflict: Transparent Knowledge Conflict Handling in Retrieval-Augmented Generation
Hua Ye, Siyuan Chen, Ziqi Zhong, Canran Xiao, Haoliang Zhang, Yuhan Wu, Fei Shen
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 TCR conflict-aware soft prompt over plain prompting | Llama-3-8B RAG · Wikidata-Conflict-5K | F1: 71.2 -> 84.3 | Improved |
| add TCR | Llama-3-8B RAG · ConflictTQA | KGRR: 42.4 -> 69.2 | Improved |
| add TCR | Llama-3-8B RAG · ConflictTQA | MCOR: 19.5 -> 61.9 | Mixed |
| add TCR | Llama-3-8B RAG · 30% distractor injection | EM drop: 18.4 -> 7.2 | Improved |
| combine TCR with RAAT | TCR · 30% distractor injection | EM drop: 7.2 -> 6.1 | Improved |
| add contrastive decoding CD2 to TCR | TCR · Wikidata-Conflict-5K | AUROC: 0.901 -> 0.905 | No clear effect |
| add TCR over Astute RAG on single-document NQ | RAG pipeline · Natural Questions (single top-1 page) | accuracy (%): 51.2 -> 54.8 | Improved |
| add self-answerability signal (ablation) | TCR · ConflictTQA/PQA ablation | MCOR/KGRR: removing SA drops ~18/26 pp | Improved |
| use SNR-based dynamic weighting instead of a fixed scalar (ablation) | TCR · ConflictTQA/PQA ablation | MCOR/KGRR: fixed degrades 17/20 pp | Improved |
| add signal integration (soft-prompt fusion) (ablation) | TCR · ConflictTQA/PQA ablation | MCOR/KGRR: removing SI collapses 55/53 pp | Improved |
| higher self-answerability bin (>0.7 vs <0.3) | TCR · flip rate analysis | flip rate: 41.3% -> 4.7% | Improved |
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.
- IRCAN — IRCAN: Mitigating Knowledge Conflicts in LLM Generation via Identifying and Reweighting Context-Aware Neurons (NeurIPS 2024) · better on 8 of 8 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
- Self-RAG: Learning to Retrieve, Generate, and Critique through Self-Reflection (ICLR 2024 · similarity 0.89)
- Adaptive Chameleon or Stubborn Sloth: Revealing the Behavior of Large Language Models in Knowledge Conflicts (ICLR 2024 · similarity 0.89)
- ParamMute: Suppressing Knowledge-Critical FFNs for Faithful Retrieval-Augmented Generation (NeurIPS 2025 · similarity 0.89)
- SePer: Measure Retrieval Utility Through The Lens Of Semantic Perplexity Reduction (ICLR 2025 · similarity 0.88)
- Retrieval from Within: An Intrinsic Capability of Attention-Based Models (NeurIPS 2026 · similarity 0.88)
- Retrieval is Not Enough: Enhancing RAG through Test-Time Critique and Optimization (NeurIPS 2025 · 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.
