Enhancing Uncertainty Modeling with Semantic Graph for Hallucination Detection

TL;DR

This paper introduces a semantic graph-based method to improve uncertainty modeling for hallucination detection in large language models, significantly enhancing detection accuracy across multiple tokens and sentences.

Contribution

It proposes a novel semantic graph approach that captures relations among tokens and sentences, enabling better uncertainty propagation and calibration for hallucination detection.

Findings

Achieved 19.78% improvement in passage-level hallucination detection

Enhanced uncertainty modeling with semantic relations among entities and sentences

Demonstrated effectiveness on two benchmark datasets

Abstract

Large Language Models (LLMs) are prone to hallucination with non-factual or unfaithful statements, which undermines the applications in real-world scenarios. Recent researches focus on uncertainty-based hallucination detection, which utilizes the output probability of LLMs for uncertainty calculation and does not rely on external knowledge or frequent sampling from LLMs. Whereas, most approaches merely consider the uncertainty of each independent token, while the intricate semantic relations among tokens and sentences are not well studied, which limits the detection of hallucination that spans over multiple tokens and sentences in the passage. In this paper, we propose a method to enhance uncertainty modeling with semantic graph for hallucination detection. Specifically, we first construct a semantic graph that well captures the relations among entity tokens and sentences. Then, we incorporate the relations between two entities for uncertainty propagation to enhance sentence-level hallucination detection. Given that hallucination occurs due to the conflict between sentences, we further present a graph-based uncertainty calibration method that integrates the contradiction probability of the sentence with its neighbors in the semantic graph for uncertainty calculation. Extensive experiments on two datasets show the great advantages of our proposed approach. In particular, we obtain substantial improvements with 19.78% in passage-level hallucination detection.