# Debiasing Large Language Models in Thai Political Stance Detection via Counterfactual Calibration

arXiv: https://arxiv.org/abs/2509.21946 · HTML: https://mlsanity.com/paper/2509.21946/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 11 of its 20 references that are in the map.
- First posted on arXiv 26 Sep 2025.

## Builds on (most cited first, 11 of 11)

- [LLaMA: Open and Efficient Foundation Language Models](https://mlsanity.com/paper/2302.13971/) (2023)
- [An Empirical Study of Catastrophic Forgetting in Large Language Models During Continual Fine-tuning](https://mlsanity.com/paper/2308.08747/) (2023)
- [Large Language Models Are Not Robust Multiple Choice Selectors](https://mlsanity.com/paper/2309.03882/) (2023)
- [Explore Spurious Correlations at the Concept Level in Language Models for Text Classification](https://mlsanity.com/paper/2311.08648/) (2023)
- [X-Stance: A Multilingual Multi-Target Dataset for Stance Detection](https://mlsanity.com/paper/2003.08385/) (2020)
- [Culture is Not Trivia: Sociocultural Theory for Cultural NLP](https://mlsanity.com/paper/2502.12057/) (2025)
- [Debiasing Algorithm through Model Adaptation](https://mlsanity.com/paper/2310.18913/) (2023)
- [Understanding the Effect of Model Compression on Social Bias in Large Language Models](https://mlsanity.com/paper/2312.05662/) (2023)
- [Co2PT: Mitigating Bias in Pre-trained Language Models through Counterfactual Contrastive Prompt Tuning](https://mlsanity.com/paper/2310.12490/) (2023)
- [Debiasing Stance Detection Models with Counterfactual Reasoning and Adversarial Bias Learning](https://mlsanity.com/paper/2212.10392/) (2022)
- [Inducing Political Bias Allows Language Models Anticipate Partisan Reactions to Controversies](https://mlsanity.com/paper/2311.09687/) (2023)

## Abstract (arXiv)

Political stance detection in low-resource and culturally complex settings poses a critical challenge for large language models (LLMs). In the Thai political landscape - marked by indirect language, polarized figures, and entangled sentiment and stance - LLMs often display systematic biases such as sentiment leakage and favoritism toward entities. These biases undermine fairness and reliability. We present ThaiFACTUAL, a lightweight, model-agnostic calibration framework that mitigates political bias without requiring fine-tuning. ThaiFACTUAL uses counterfactual data augmentation and rationale-based supervision to disentangle sentiment from stance and reduce bias. We also release the first high-quality Thai political stance dataset, annotated with stance, sentiment, rationales, and bias markers across diverse entities and events. Experimental results show that ThaiFACTUAL significantly reduces spurious correlations, enhances zero-shot generalization, and improves fairness across multiple LLMs. This work highlights the importance of culturally grounded debiasing techniques for underrepresented languages.