SINdex: Semantic INconsistency Index for Hallucination Detection in LLMs

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Abdaljalil, Samir, Kurban, Hasan, Sharma, Parichit, Serpedin, Erchin, Atat, Rachad
Formato: Preprint
Publicado: 2025
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866909531657732096
author Abdaljalil, Samir
Kurban, Hasan
Sharma, Parichit
Serpedin, Erchin
Atat, Rachad
author_facet Abdaljalil, Samir
Kurban, Hasan
Sharma, Parichit
Serpedin, Erchin
Atat, Rachad
contents Large language models (LLMs) are increasingly deployed across diverse domains, yet they are prone to generating factually incorrect outputs - commonly known as "hallucinations." Among existing mitigation strategies, uncertainty-based methods are particularly attractive due to their ease of implementation, independence from external data, and compatibility with standard LLMs. In this work, we introduce a novel and scalable uncertainty-based semantic clustering framework for automated hallucination detection. Our approach leverages sentence embeddings and hierarchical clustering alongside a newly proposed inconsistency measure, SINdex, to yield more homogeneous clusters and more accurate detection of hallucination phenomena across various LLMs. Evaluations on prominent open- and closed-book QA datasets demonstrate that our method achieves AUROC improvements of up to 9.3% over state-of-the-art techniques. Extensive ablation studies further validate the effectiveness of each component in our framework.
format Preprint
id arxiv_https___arxiv_org_abs_2503_05980
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle SINdex: Semantic INconsistency Index for Hallucination Detection in LLMs
Abdaljalil, Samir
Kurban, Hasan
Sharma, Parichit
Serpedin, Erchin
Atat, Rachad
Computation and Language
Artificial Intelligence
Large language models (LLMs) are increasingly deployed across diverse domains, yet they are prone to generating factually incorrect outputs - commonly known as "hallucinations." Among existing mitigation strategies, uncertainty-based methods are particularly attractive due to their ease of implementation, independence from external data, and compatibility with standard LLMs. In this work, we introduce a novel and scalable uncertainty-based semantic clustering framework for automated hallucination detection. Our approach leverages sentence embeddings and hierarchical clustering alongside a newly proposed inconsistency measure, SINdex, to yield more homogeneous clusters and more accurate detection of hallucination phenomena across various LLMs. Evaluations on prominent open- and closed-book QA datasets demonstrate that our method achieves AUROC improvements of up to 9.3% over state-of-the-art techniques. Extensive ablation studies further validate the effectiveness of each component in our framework.
title SINdex: Semantic INconsistency Index for Hallucination Detection in LLMs
topic Computation and Language
Artificial Intelligence
url https://arxiv.org/abs/2503.05980