dorsal/arxiv
View SchemaSeeing through the Conflict: Transparent Knowledge Conflict Handling in Retrieval-Augmented Generation
| Authors | Hua Ye, Siyuan Chen, Ziqi Zhong, Canran Xiao, Haoliang Zhang, Yuhan Wu, Fei Shen |
|---|---|
| Categories | |
| ArXiv ID | 2601.06842vv1 |
| URL | https://arxiv.org/abs/2601.06842 |
| License | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ |
Abstract
Large language models (LLMs) equipped with retrieval--the Retrieval-Augmented Generation (RAG) paradigm--should combine their parametric knowledge with external evidence, yet in practice they often hallucinate, over-trust noisy snippets, or ignore vital context. We introduce TCR (Transparent Conflict Resolution), a plug-and-play framework that makes this decision process observable and controllable. TCR (i) disentangles semantic match and factual consistency via dual contrastive encoders, (ii) estimates self-answerability to gauge confidence in internal memory, and (iii) feeds the three scalar signals to the generator through a lightweight soft-prompt with SNR-based weighting. Across seven benchmarks TCR improves conflict detection (+5-18 F1), raises knowledge-gap recovery by +21.4 pp and cuts misleading-context overrides by -29.3 pp, while adding only 0.3% parameters. The signals align with human judgements and expose temporal decision patterns.
{
"annotation_id": "b8fe6ddc-2446-495d-b6e1-07c8b7efd8b3",
"date_created": "2026-02-17T05:53:08.863000Z",
"date_modified": "2026-02-17T05:53:08.863000Z",
"file_hash": "c17d20dd5734b32339b1d43341d40980389a9096e135ec2b54a5b5963099c1dc",
"private": false,
"record": {
"abstract": "Large language models (LLMs) equipped with retrieval--the Retrieval-Augmented Generation (RAG) paradigm--should combine their parametric knowledge with external evidence, yet in practice they often hallucinate, over-trust noisy snippets, or ignore vital context. We introduce TCR (Transparent Conflict Resolution), a plug-and-play framework that makes this decision process observable and controllable. TCR (i) disentangles semantic match and factual consistency via dual contrastive encoders, (ii) estimates self-answerability to gauge confidence in internal memory, and (iii) feeds the three scalar signals to the generator through a lightweight soft-prompt with SNR-based weighting. Across seven benchmarks TCR improves conflict detection (+5-18 F1), raises knowledge-gap recovery by +21.4 pp and cuts misleading-context overrides by -29.3 pp, while adding only 0.3% parameters. The signals align with human judgements and expose temporal decision patterns.",
"arxiv_id": "2601.06842",
"authors": [
"Hua Ye",
"Siyuan Chen",
"Ziqi Zhong",
"Canran Xiao",
"Haoliang Zhang",
"Yuhan Wu",
"Fei Shen"
],
"categories": [
"cs.AI"
],
"license": "http://arxiv.org/licenses/nonexclusive-distrib/1.0/",
"title": "Seeing through the Conflict: Transparent Knowledge Conflict Handling in Retrieval-Augmented Generation",
"url": "https://arxiv.org/abs/2601.06842",
"version": "v1"
},
"schema_id": "dorsal/arxiv",
"source": {
"execution_id": "f0855221-50c5-447c-96cd-85116c2d2471",
"id": "arXiv Dataset",
"type": "Model",
"variant": "snapshot-2026-01-17",
"version": "0.1.0"
},
"user_id": 1000002
}