Beyond Semantic Relevance: Counterfactual Risk Minimization for Robust Retrieval-Augmented Generation

TL;DR

This paper introduces CoRM-RAG, a retrieval framework that prioritizes decision safety over semantic relevance, improving robustness against user biases and adversarial queries in retrieval-augmented generation systems.

Contribution

It proposes a causal intervention-based training protocol and a lightweight Evidence Critic to enhance retrieval robustness and risk-awareness in RAG systems.

Findings

CoRM-RAG outperforms existing retrievers in adversarial settings.

The Evidence Critic effectively identifies evidentially strong documents.

The framework enables risk-aware abstention for safer decision-making.

Abstract

Standard Retrieval-Augmented Generation (RAG) systems predominantly rely on semantic relevance as a proxy for utility. However, this assumption collapses in realistic decision-making scenarios where user queries are laden with cognitive biases, such as false premises or confirmation bias. In such cases, maximizing relevance paradoxically promotes the retrieval of sycophantic evidence that reinforces hallucinations, a critical failure we term the ``Relevance-Robustness Gap''. To bridge this gap, we propose CoRM-RAG (Counterfactual Risk Minimization for RAG), a framework that aligns retrieval with decision safety rather than mere similarity. Grounded in causal intervention, we introduce a Cognitive Perturbation Protocol to simulate user biases during training, which is then distilled into a lightweight Evidence Critic. This scoring module learns to identify documents that possess sufficient evidential strength to steer the model toward correctness despite adversarial query perturbations. Extensive experiments on decision-making benchmarks demonstrate that CoRM-RAG significantly outperforms strong dense retrievers and LLM-based rerankers in adversarial settings, while enabling effective risk-aware abstention through reliable robustness scoring. Our code is available at https://github.com/PeiYangLiu/CoRM-RAG.git.