🧠 BiasAtlas — an interactive atlas of AI bias research

Live atlas: https://biasatlas.cair-nepal.org.  Companion paper: Whose fairness? Structural concentration in AI bias research

BiasAtlas is an interactive, continuously updatable map of the AI bias research landscape. It lets anyone explore who produces research on AI bias, where, in collaboration with whom, and on which themes — and to track how that structure shifts over time. The atlas accompanies our study of 692 publications (2012–2026) and turns a static bibliometric snapshot into a living resource the community can browse, query, and contribute to.

This work is part of Bridging Minds and Machines: Human Perspectives and Responsible AI for an Inclusive Future project.

Authors: Abhash Shrestha, Subigya Gautam, Anu Sapkota, Sanju Tiwari and Tek Raj Chhetri

---

What you can explore

• Semantic map of the field — a UMAP projection of Sentence-BERT abstract embeddings, coloured by thematic domain, so related work sits together.
• Five thematic domains — General Fairness & Bias Mitigation, Health & Clinical AI, LLMs & NLP, Recommender Systems, and Graph-Based Fairness & Bias Mitigation.
• Geographic & institutional structure — country- and institution-level contributions, distinguishing participation (all-author) from leadership (first-author).
• Collaboration network — co-authorship structure and within- versus cross-region collaboration.
• Citation dynamics — within-corpus and global (OpenAlex) citation influence per domain and per paper.
• Temporal evolution — how volume and thematic focus have changed year on year.
• Continuously growing corpus — new publications are added over time to keep the atlas current (see Contributing to nominate one).

📄 Citation

If you use BiasAtlas or the underlying corpus, please cite the paper:

@misc{shrestha2026fairnessstructuralconcentrationai,
      title={Whose fairness? Structural concentration in AI bias research}, 
      author={Abhash Shrestha and Subigya Gautam and Anu Sapkota and Sanju Tiwari and Tek Raj Chhetri},
      year={2026},
      eprint={2607.05574},
      archivePrefix={arXiv},
      primaryClass={cs.CY},
      url={https://arxiv.org/abs/2607.05574}, 
}
}

To cite the dataset itself (versioned, citable independently of the paper), use the archived Zenodo record: https://doi.org/10.5281/zenodo.21221283. The corpus is also available on Hugging Face: https://huggingface.co/datasets/cair-nepal/ai-bias-research-landscape

Authors

• Abhash Shrestha — Center for AI Research (CAIR) Nepal (corresponding)
• Subigya Gautam — Center for AI Research (CAIR) Nepal
• Anu Sapkota — Center for AI Research (CAIR) Nepal
• Sanju Tiwari — Sharda University, Delhi-NCR; Shodhguru Innovation and Research Labs
• Tek Raj Chhetri — Center for AI Research (CAIR) Nepal; McGovern Institute for Brain Research, MIT (corresponding | Supervisor)

Contact: abhash.shrestha@cair-nepal.org · tekraj.chhetri@cair-nepal.org

Contributing

Contributions are welcome — to nominate a publication for the corpus, open an issue with the paper's title, DOI, and thematic domain. To add papers yourself and open a PR, or to contribute code/pipeline changes, see CONTRIBUTING.md for the step-by-step process and common pitfalls.

📜 License

This project is licensed under the Apache License 2.0.