Saved in:
Bibliographic Details
Main Authors: Vangala, Bhanu Prakash, Mahmud, Sajid, Neupane, Pawan, Selvaraj, Joel, Cheng, Jianlin
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2512.22396
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909976740495360
author Vangala, Bhanu Prakash
Mahmud, Sajid
Neupane, Pawan
Selvaraj, Joel
Cheng, Jianlin
author_facet Vangala, Bhanu Prakash
Mahmud, Sajid
Neupane, Pawan
Selvaraj, Joel
Cheng, Jianlin
contents Artificial Intelligence (AI), particularly Large Language Models (LLMs), is transforming scientific discovery, enabling rapid knowledge generation and hypothesis formulation. However, a critical challenge is hallucination, where LLMs generate factually incorrect or misleading information, compromising research integrity. To address this, we introduce HalluMatData, a benchmark dataset for evaluating hallucination detection methods, factual consistency, and response robustness in AI-generated materials science content. Alongside this, we propose HalluMatDetector, a multi-stage hallucination detection framework that integrates intrinsic verification, multi-source retrieval, contradiction graph analysis, and metric-based assessment to detect and mitigate LLM hallucinations. Our findings reveal that hallucination levels vary significantly across materials science subdomains, with high-entropy queries exhibiting greater factual inconsistencies. By utilizing HalluMatDetector verification pipeline, we reduce hallucination rates by 30% compared to standard LLM outputs. Furthermore, we introduce the Paraphrased Hallucination Consistency Score (PHCS) to quantify inconsistencies in LLM responses across semantically equivalent queries, offering deeper insights into model reliability.
format Preprint
id arxiv_https___arxiv_org_abs_2512_22396
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle HalluMat: Detecting Hallucinations in LLM-Generated Materials Science Content Through Multi-Stage Verification
Vangala, Bhanu Prakash
Mahmud, Sajid
Neupane, Pawan
Selvaraj, Joel
Cheng, Jianlin
Artificial Intelligence
Materials Science
Information Retrieval
Artificial Intelligence (AI), particularly Large Language Models (LLMs), is transforming scientific discovery, enabling rapid knowledge generation and hypothesis formulation. However, a critical challenge is hallucination, where LLMs generate factually incorrect or misleading information, compromising research integrity. To address this, we introduce HalluMatData, a benchmark dataset for evaluating hallucination detection methods, factual consistency, and response robustness in AI-generated materials science content. Alongside this, we propose HalluMatDetector, a multi-stage hallucination detection framework that integrates intrinsic verification, multi-source retrieval, contradiction graph analysis, and metric-based assessment to detect and mitigate LLM hallucinations. Our findings reveal that hallucination levels vary significantly across materials science subdomains, with high-entropy queries exhibiting greater factual inconsistencies. By utilizing HalluMatDetector verification pipeline, we reduce hallucination rates by 30% compared to standard LLM outputs. Furthermore, we introduce the Paraphrased Hallucination Consistency Score (PHCS) to quantify inconsistencies in LLM responses across semantically equivalent queries, offering deeper insights into model reliability.
title HalluMat: Detecting Hallucinations in LLM-Generated Materials Science Content Through Multi-Stage Verification
topic Artificial Intelligence
Materials Science
Information Retrieval
url https://arxiv.org/abs/2512.22396