Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

FinVet: A Collaborative Framework of RAG and External Fact-Checking Agents for Financial Misinformation Detection

Daniel Berhane Araya, Duoduo Liao

TL;DR

This paper tackles the challenge of financial misinformation by introducing FinVet, a multi-agent framework that fuses two Retrieval-Augmented Generation (RAG) pipelines with an external fact-checking module, all coordinated via a confidence-weighted voting mechanism. An adaptive three-tier evidence processing strategy uses retrieval similarity to switch among direct metadata extraction, hybrid reasoning, and model-based analysis, delivering transparent verdicts with source attribution and quantified uncertainty. The framework is validated on the FinFact dataset, achieving an F1 score of 0.85, a substantial improvement over both single pipelines and standalone RAG approaches, and demonstrating the benefits of collaborative verification for explainability in finance. These results suggest FinVet’s approach provides robust, auditable detection of financial misinformation with practical implications for institutions and regulators managing market integrity and investor protection.

Abstract

Financial markets face growing threats from misinformation that can trigger billions in losses in minutes. Most existing approaches lack transparency in their decision-making and provide limited attribution to credible sources. We introduce FinVet, a novel multi-agent framework that integrates two Retrieval-Augmented Generation (RAG) pipelines with external fact-checking through a confidence-weighted voting mechanism. FinVet employs adaptive three-tier processing that dynamically adjusts verification strategies based on retrieval confidence, from direct metadata extraction to hybrid reasoning to full model-based analysis. Unlike existing methods, FinVet provides evidence-backed verdicts, source attribution, confidence scores, and explicit uncertainty flags when evidence is insufficient. Experimental evaluation on the FinFact dataset shows that FinVet achieves an F1 score of 0.85, which is a 10.4% improvement over the best individual pipeline (fact-check pipeline) and 37% improvement over standalone RAG approaches.

FinVet: A Collaborative Framework of RAG and External Fact-Checking Agents for Financial Misinformation Detection

TL;DR

This paper tackles the challenge of financial misinformation by introducing FinVet, a multi-agent framework that fuses two Retrieval-Augmented Generation (RAG) pipelines with an external fact-checking module, all coordinated via a confidence-weighted voting mechanism. An adaptive three-tier evidence processing strategy uses retrieval similarity to switch among direct metadata extraction, hybrid reasoning, and model-based analysis, delivering transparent verdicts with source attribution and quantified uncertainty. The framework is validated on the FinFact dataset, achieving an F1 score of 0.85, a substantial improvement over both single pipelines and standalone RAG approaches, and demonstrating the benefits of collaborative verification for explainability in finance. These results suggest FinVet’s approach provides robust, auditable detection of financial misinformation with practical implications for institutions and regulators managing market integrity and investor protection.

Abstract

Financial markets face growing threats from misinformation that can trigger billions in losses in minutes. Most existing approaches lack transparency in their decision-making and provide limited attribution to credible sources. We introduce FinVet, a novel multi-agent framework that integrates two Retrieval-Augmented Generation (RAG) pipelines with external fact-checking through a confidence-weighted voting mechanism. FinVet employs adaptive three-tier processing that dynamically adjusts verification strategies based on retrieval confidence, from direct metadata extraction to hybrid reasoning to full model-based analysis. Unlike existing methods, FinVet provides evidence-backed verdicts, source attribution, confidence scores, and explicit uncertainty flags when evidence is insufficient. Experimental evaluation on the FinFact dataset shows that FinVet achieves an F1 score of 0.85, which is a 10.4% improvement over the best individual pipeline (fact-check pipeline) and 37% improvement over standalone RAG approaches.

Paper Structure

This paper contains 24 sections, 1 figure, 3 tables, 2 algorithms.

Table of Contents

  1. Introduction
  2. Related Work
  3. Deep Learning-Based Techniques
  4. LLM-based Approaches
  5. Summary of Contributions
  6. Methodology
  7. Data Processing and Vector Store
  8. Claim Verification Pipelines
  9. RAG Verification Pipelines
  10. Fact-Check Pipeline
  11. Results Normalization
  12. Verdict Integration and Reporting
  13. Experimental Results and Discussion
  14. Experimental Setup
  15. Vector Embedding and Storage
  16. ...and 9 more sections

Figures (1)

  • Figure 1: FinVet: A Collaborative Framework of RAG and External Fact-Checking Agents for Financial Misinformation Detection