# Semantic and Structural Analysis of Implicit Biases in Large Language Models: An Interpretable Approach

arXiv: https://arxiv.org/abs/2508.06155 · HTML: https://mlsanity.com/paper/2508.06155/

## Key facts

- Cited by 5 papers in the map (data through 14 Oct 2026).
- Builds on 13 of its 27 references that are in the map.
- First posted on arXiv 8 Aug 2025.

## Builds on (most cited first, 13 of 13)

- [StereoSet: Measuring stereotypical bias in pretrained language models](https://mlsanity.com/paper/2004.09456/) (2020)
- [On Measuring Social Biases in Sentence Encoders](https://mlsanity.com/paper/1903.10561/) (2019)
- [Gender bias and stereotypes in Large Language Models](https://mlsanity.com/paper/2308.14921/) (2023)
- [UniBias: Unveiling and Mitigating LLM Bias through Internal Attention and FFN Manipulation](https://mlsanity.com/paper/2405.20612/) (2024)
- [Towards Auditing Large Language Models: Improving Text-based Stereotype Detection](https://mlsanity.com/paper/2311.14126/) (2023)
- [Fairness Mediator: Neutralize Stereotype Associations to Mitigate Bias in Large Language Models](https://mlsanity.com/paper/2504.07787/) (2025)
- [Detecting Linguistic Indicators for Stereotype Assessment with Large Language Models](https://mlsanity.com/paper/2502.19160/) (2025)
- [Capsule Network-Based Semantic Intent Modeling for Human-Computer Interaction](https://mlsanity.com/paper/2507.00540/) (2025)
- [Collaborative Distillation Strategies for Parameter-Efficient Language Model Deployment](https://mlsanity.com/paper/2507.15198/) (2025)
- [Towards Robust Few-Shot Text Classification Using Transformer Architectures and Dual Loss Strategies](https://mlsanity.com/paper/2505.06145/) (2025)
- [Structured Gradient Guidance for Few-Shot Adaptation in Large Language Models](https://mlsanity.com/paper/2506.00726/) (2025)
- [StereoDetect: Detecting Stereotypes and Anti-stereotypes the Correct Way Using Social Psychological Underpinnings](https://mlsanity.com/paper/2504.03352/) (2025)
- [FairMonitor: A Four-Stage Automatic Framework for Detecting Stereotypes and Biases in Large Language Models](https://mlsanity.com/paper/2308.10397/) (2023)

## Built on by (most cited first, 5 of 5)

- [Trustworthy Summarization via Uncertainty Quantification and Risk Awareness in Large Language Models](https://mlsanity.com/paper/2510.01231/) (2025)
- [Modular Task Decomposition and Dynamic Collaboration in Multi-Agent Systems Driven by Large Language Models](https://mlsanity.com/paper/2511.01149/) (2025)
- [Dynamic Topic Evolution with Temporal Decay and Attention in Large Language Models](https://mlsanity.com/paper/2510.10613/) (2025)
- [LLM-Centric RAG with Multi-Granular Indexing and Confidence Constraints](https://mlsanity.com/paper/2510.27054/) (2025)
- [Topology-Aware Graph Reinforcement Learning for Dynamic Routing in Cloud Networks](https://mlsanity.com/paper/2509.04973/) (2025)

## Abstract (arXiv)

This paper addresses the issue of implicit stereotypes that may arise during the generation process of large language models. It proposes an interpretable bias detection method aimed at identifying hidden social biases in model outputs, especially those semantic tendencies that are not easily captured through explicit linguistic features. The method combines nested semantic representation with a contextual contrast mechanism. It extracts latent bias features from the vector space structure of model outputs. Using attention weight perturbation, it analyzes the model's sensitivity to specific social attribute terms, thereby revealing the semantic pathways through which bias is formed. To validate the effectiveness of the method, this study uses the StereoSet dataset, which covers multiple stereotype dimensions including gender, profession, religion, and race. The evaluation focuses on several key metrics, such as bias detection accuracy, semantic consistency, and contextual sensitivity. Experimental results show that the proposed method achieves strong detection performance across various dimensions. It can accurately identify bias differences between semantically similar texts while maintaining high semantic alignment and output stability. The method also demonstrates high interpretability in its structural design. It helps uncover the internal bias association mechanisms within language models. This provides a more transparent and reliable technical foundation for bias detection. The approach is suitable for real-world applications where high trustworthiness of generated content is required.