My research lies at the intersection of Natural Language Processing (NLP), Computational Social Science, and Culturally Adaptive NLP applications. I am particularly interested in understanding how language models can better reflect and respect human culture.
My research interests are:
Cultural Understanding in Language Models: As autoregressive generators, can modern language models truly perceive and represent culture, which evolves over thousands of years, as humans do?
Culturally Adapted NLP: My recent work explores how Large Language Models (LLMs) can be adapted to culturally sensitive domains such as mental health, irony, and sarcasm, where context and cultural nuance play a crucial role.
Cultural Bias: I am also interested in understanding how cultural bias is manifested and propagated in language models. I approach this through the lens of interpretability to uncover its roots and explore ways to mitigate it.
I aim to build NLP systems that go beyond linguistic accuracy, incorporating cultural awareness and empathy into machine understanding of human language.
🎉 Our work on Multilingual Financial Misinformation is accepted at ACL Findings 2026. Available on arXiv.
Feb 13, 2026
🎉 Our collaborative work on Cross-Cultural Translation is accepted at LREC 2026. Available on arXiv.
Jan 23, 2026
🎉 Our work on Islamic lifestyle applications bas been accepted by the International Journal of Human–Computer Interaction . The paper is open-access and available here.
Oct 24, 2025
🎉 Our work on Religious Bias bas been accepted by the Jounral of AI & Society . The paper is available on arXiv.
@article{kabir2025checkbox,author={Kabir, M and Abrar, A and Ananiadou, S},journal={Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing},year={2025},month=nov,note={https://aclanthology.org/2025.emnlp-main.2/},publisher={Association for Computational Linguistics},title={Break the Checkbox: Challenging Closed-Style Evaluations of Cultural Alignment in {LLMs}},doi={10.18653/v1/2025.emnlp-main.2},}
NLP
From n-gram to Attention: How Model Architectures Learn and Propagate Bias in Language Modeling
M Kabir, T Tahsin, and S Ananiadou
Findings of the Association for Computational Linguistics: EMNLP 2025, Nov 2025
@article{kabir2025attention,author={Kabir, M and Tahsin, T and Ananiadou, S},journal={Findings of the Association for Computational Linguistics: EMNLP 2025},year={2025},month=nov,note={https://aclanthology.org/2025.findings-emnlp.1003/},publisher={Association for Computational Linguistics},title={From n-gram to Attention: How Model Architectures Learn and Propagate Bias in Language Modeling},doi={10.18653/v1/2025.findings-emnlp.1003},}
NLP
Religious Bias Landscape in Language and Text-to-Image Models: Analysis, Detection, and Debiasing Strategies
Ajwad Abrar, Nafisa Tabassum Oeshy, Mohsinul Kabir, and 1 more author
@article{abrar2025religious,author={Abrar, Ajwad and Oeshy, Nafisa Tabassum and Kabir, Mohsinul and Ananiadou, Sophia},journal={AI \& SOCIETY},year={2025},month=nov,note={Published online},publisher={Springer},title={Religious Bias Landscape in Language and Text-to-Image Models: Analysis, Detection, and Debiasing Strategies},doi={10.1007/s00146-025-02721-z},}
NLP
Semantic Label Drift in Cross-Cultural Translation
Mohsinul Kabir, Tasnim Ahmed, Md Mezbaur Rahman, and 2 more authors
@article{kabir2025semantic,author={Kabir, Mohsinul and Ahmed, Tasnim and Rahman, Md Mezbaur and Giannouris, Polydoros and Ananiadou, Sophia},journal={arXiv},year={2025},month=oct,note={https://arxiv.org/abs/2510.25967},publisher={},title={Semantic Label Drift in Cross-Cultural Translation},doi={10.48550/arXiv.2510.25967},}
