LLM4Trust: Exploring the Capabilities of Large Language Models for Trust Evaluation

2026-09-27

AI Summary

This study addresses the limitations of traditional trust assessment, which relies heavily on extensive annotations and lacks interpretability, by pioneering a systematic trust evaluation framework based on large language models (LLMs). Methodologically, this work constructs the first LLM-based trust assessment benchmark, integrating prompt engineering with trust graph modeling. The zero- and few-shot reasoning capabilities of LLMs are validated through multi-task evaluations, while graph information extraction and adversarial defense mechanisms are proposed to optimize batch inference efficiency. Empirical results demonstrate that LLMs possess effective trust assessment potential under weakly supervised conditions. Furthermore, the proposed strategies significantly enhance system robustness and interpretability, establishing a novel paradigm for intelligent trust computing.

Abstract

Trust evaluation plays a critical role in cybersecurity by supporting risk mitigation and decision-making. A variety of trust evaluation methods have been proposed, with learning-based approaches offering high accuracy and automation. However, they often require substantial ground truth, suffer from low training efficiency, lack support for basic trust properties, and provide limited explainability. Large Language Models (LLMs) offer a compelling alternative due to their strong zero-/few-shot reasoning abilities and broad knowledge. To this end, we propose LLM4Trust, the first benchmark framework that systematically explores the capabilities of LLMs for trust evaluation. We first construct diverse trust graphs to model five basic trust properties and design corresponding property understanding tasks. We then assess the ability of eight representative LLMs to understand these properties under nine prompt methods. Based on this exploration, we identify the most effective LLM-prompt combinations and apply them to five real-world datasets for validating LLMs'trust evaluation capability. During this process, we propose two strategies to extract key information from large-scale trust graphs, addressing the context window limitations of LLMs. Extensive experiments show that LLMs can effectively understand basic trust properties and have great potential for real-world trust evaluation, particularly under limited supervision. However, they remain vulnerable to attacks targeting trust graphs and demonstration examples used in few-shot prompting, and incur high inference costs. Accordingly, we propose a defense mechanism and batch inference to improve the robustness and efficiency of LLM-based trust evaluation. The source code of LLM4Trust is available at https://github.com/Jieerbobo/LLM4Trust

Problem

• Trust Evaluation
• Large Language Models
• Cybersecurity
• Benchmark
• Robustness

Innovation

• Large Language Models
• Trust Evaluation
• Benchmark Framework
• Prompt Engineering
• Robustness Defense