Mitigating Hallucinations in Finance-Based Multi-Agent Large Language Model Systems
Name
electronics-15-04456.pdf
Size
1.05 MB
Format
Adobe PDF
Checksum (MD5)
561c7a44b2a6e2ef38b3ddc4e6cfd584
Author(s)
Nagpal, Rashmi
Usua, Unyimeabasi
Simons, Kailey
Queipo, Sabrina
Gupta, Amar
Gossain, Vishal
Date Issued
September 28, 2026
Journal
Electronics
Publisher
Multidisciplinary Digital Publishing Institute
Citation
Nagpal, R.; Usua, U.; Simons, K.; Queipo, S.; Gupta, A.; Gossain, V. Mitigating Hallucinations in Finance-Based Multi-Agent Large Language Model Systems. Electronics 2026, 15, 4456.
Version
Final published version
Abstract
Large language models are increasingly used for financial question answering, while they are prone to generating hallucinated content. In this research, we propose a multi-signal framework for hallucination detection and mitigation. Our framework combines six signals (entailment, semantic similarity, claim verification, numeric consistency, token overlap and confidence calibration techniques) to identify hallucination type and severity. We evaluated our framework using multiple language models across three financial QA benchmark datasets (FinQA, FinanceQA, and FinDER). Our proposed framework corrects up to 78.4% of the detected hallucinations (fix rate on the hallucinated subset of FinQA; 47.7% on FinanceQA and 45.7% on FinDER), outperforming a type-agnostic mitigation baseline on all three datasets. This highlights that targeted, type-aware hallucination mitigation can significantly improve answer reliability while remaining computationally effective.
MIT Department
Massachusetts Institute of Technology. Computer Science and Artificial Intelligence Laboratory
Terms of Use
Creative Commons Attribution
Persistent DSpace Link
DOI of Published Version
https://doi.org/10.3390/electronics15194456
