FaithfulRAG: Fact-Level Conflict Modeling for Context-Faithful Retrieval-Augmented Generation

2025-06-10

AI Summary

In RAG systems, conflicts between LLMs’ parametric knowledge and retrieved context frequently cause unfaithful generation—either ignoring or erroneously blending contextual information. Existing approaches enforce faithfulness by suppressing parametric knowledge, but this compromises the model’s intrinsic knowledge structure and contextual comprehension. This paper proposes FACT-RAG: a fact-grained framework that explicitly models knowledge discrepancies between parametric and contextual sources. It introduces a self-reflective reasoning architecture comprising (i) a fact alignment and conflict identification module, and (ii) a dynamic knowledge fusion decoding strategy that guides the LLM to actively detect, analyze, and reconcile conflicts before generation. Unlike suppression-based methods, FACT-RAG preserves both contextual faithfulness and parametric knowledge utility. Extensive evaluation on multiple knowledge-intensive benchmarks demonstrates significant improvements over state-of-the-art methods in answer faithfulness and factual accuracy. The code is publicly available.

Technology Category

• Natural Language Processing: Fact-Checking / Misinformation Detection (NLP Focus)
• Knowledge Representation and Reasoning: Reasoning with Beliefs
• Machine Learning: Large Multimodal Models (LMMs)

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

Large language models (LLMs) augmented with retrieval systems have demonstrated significant potential in handling knowledge-intensive tasks. However, these models often struggle with unfaithfulness issues, generating outputs that either ignore the retrieved context or inconsistently blend it with the LLM`s parametric knowledge. This issue is particularly severe in cases of knowledge conflict, where the retrieved context conflicts with the model`s parametric knowledge. While existing faithful RAG approaches enforce strict context adherence through well-designed prompts or modified decoding strategies, our analysis reveals a critical limitation: they achieve faithfulness by forcibly suppressing the model`s parametric knowledge, which undermines the model`s internal knowledge structure and increases the risk of misinterpreting the context. To this end, this paper proposes FaithfulRAG, a novel framework that resolves knowledge conflicts by explicitly modeling discrepancies between the model`s parametric knowledge and retrieved context. Specifically, FaithfulRAG identifies conflicting knowledge at the fact level and designs a self-thinking process, allowing LLMs to reason about and integrate conflicting facts before generating responses. Extensive experiments demonstrate that our method outperforms state-of-the-art methods. The code is available at https:// github.com/DeepLearnXMU/Faithful-RAG

Problem

• Resolves knowledge conflicts in retrieval-augmented generation
• Models discrepancies between parametric and retrieved knowledge
• Ensures faithful integration of conflicting facts in responses

Innovation

• Models fact-level conflicts explicitly
• Integrates self-thinking for reasoning
• Balances context and parametric knowledge