Mitigating Hallucination in Financial Retrieval-Augmented Generation via Fine-Grained Knowledge Verification

TL;DR

This paper introduces a reinforcement learning approach with fine-grained knowledge verification to reduce hallucinations in financial retrieval-augmented generation systems, leading to more accurate and faithful responses.

Contribution

It proposes a novel reinforcement learning framework that decomposes responses into knowledge units and assesses their correctness to improve alignment with retrieved documents in financial NLP tasks.

Findings

Significant reduction in hallucinations on FDD and FDD-ANT datasets.

Improved response fidelity and informativeness compared to baseline models.

Enhanced alignment with retrieved financial documents.

Abstract

In financial Retrieval-Augmented Generation (RAG) systems, models frequently rely on retrieved documents to generate accurate responses due to the time-sensitive nature of the financial domain. While retrieved documents help address knowledge gaps, model-generated responses still suffer from hallucinations that contradict the retrieved information. To mitigate this inconsistency, we propose a Reinforcement Learning framework enhanced with Fine-grained Knowledge Verification (RLFKV). Our method decomposes financial responses into atomic knowledge units and assesses the correctness of each unit to compute the fine-grained faithful reward. This reward offers more precise optimization signals, thereby improving alignment with the retrieved documents. Additionally, to prevent reward hacking (e.g., overly concise replies), we incorporate an informativeness reward that encourages the policy model to retain at least as many knowledge units as the base model. Experiments conducted on the public Financial Data Description (FDD) task and our newly proposed FDD-ANT dataset demonstrate consistent improvements, confirming the effectiveness of our approach.