CQC-RAG: Robust Retrieval-Augmented Generation via Cross-Query Consistency

2026-06-11

AI Summary

This work addresses the vulnerability of Retrieval-Augmented Generation (RAG) to query formulation, where variations in phrasing can yield divergent retrieval results or introduce irrelevant documents, thereby inducing hallucinations and undermining answer reliability. The authors propose a cross-query consistency hypothesis, leveraging semantically equivalent yet syntactically diverse query rewrites to construct a shared document pool, upon which conditional reasoning is performed. They further exploit the stability of answer confidence across different rewritten contexts to enable self-evaluation and selection. This approach innovatively integrates query diversity injection with consistency-based assessment into a unified, supervision-free hallucination filtering mechanism. Evaluated on four open-domain question answering benchmarks, the method significantly outperforms state-of-the-art multi-query approaches, achieving absolute exact-match improvements of 4.76 and 9.12 percentage points on TriviaQA and MuSiQue, respectively.

Abstract

Retrieval-Augmented Generation (RAG) has become a common approach for improving the factuality of Large Language Models (LLMs), yet its reliability remains highly sensitive to how external evidence is retrieved and used. Semantically equivalent queries with different syntactic forms may lead to different retrieval results, while irrelevant or misleading documents can further induce hallucinated answers. Existing multi-path reasoning methods improve robustness by sampling multiple candidate answers and applying voting- or confidence-based selection, but they still face two limitations: diversity is often injected through uncontrollable decoding randomness, and answer evaluation is usually confined to a single query-induced evidence view. To address these limitations, we propose a Cross-Query Consistency Hypothesis: correct answers tend to maintain high confidence across semantically equivalent but syntactically diverse queries, whereas noise-induced hallucinations exhibit unstable confidence under such query variations. Based on this hypothesis, we introduce CQC-RAG, a framework that co-designs query-level diversity injection with cross-query consistency evaluation. CQC-RAG rewrites the original question into diverse but meaning-preserving queries, reranks a shared document pool to construct query-conditioned reasoning contexts, applies an evidence-grounded protocol to extract answer-evidence pairs and selects answers according to their confidence stability across these contexts. This design enables self-evaluation without external supervision and does not rely on expanded retrieval coverage. Experiments on four open-domain question answering benchmarks show that CQC-RAG outperforms the strongest previous multi-query baseline by +4.76 pp EM on TriviaQA and +9.12 pp EM on MuSiQue, validating the effectiveness of cross-query consistency for filtering noise-induced hallucinations.

Problem

• Retrieval-Augmented Generation
• Hallucination
• Query Variability
• Factuality
• Robustness

Innovation

• Cross-Query Consistency
• Retrieval-Augmented Generation
• Hallucination Filtering
• Query Rewriting
• Confidence Stability