Fairness in NLP with Graph-based Representations: A Systematic Survey

ACL ARR 2026 January Submission6070 Authors

Keywords: Fairness in NLP, Graph-based representation, Social bias, Information extraction, Survey

Abstract: As large language models (LLMs) increasingly permeate diverse domains, concerns about their trustworthiness and fairness have become central ethical issues. Since social bias is highly context-dependent, understanding relational structure beyond isolated text is crucial. Graph-based representations capture structural relationships within and across texts and have recently been integrated with LLMs to address fairness with promising results. However, there is still no comprehensive review of how graphs and LLMs are jointly used for fairness. To address this gap, we provide a systematic analysis of graph-based approaches to fairness and their integration with LLMs, and outline key future research directions.

Paper Type: Long

Research Area: Ethics, Bias, and Fairness

Research Area Keywords: Ethics, Bias, and Fairness, Information Extraction, NLP for Social Good, Interpretability and Analysis of Models for NLP, Machine Learning for NLP

Contribution Types: Surveys

Languages Studied: English

Submission Number: 6070