uzw/dynamic_gender_label_dataset

This is the official dataset for Beyond Binary Gender Labels: Revealing Gender Bias in LLMs through Gender-Neutral Name Predictions

Name-based gender prediction has traditionally categorized individuals as either female or male based on their names, using a binary classification system. That binary approach can be problematic in the cases of gender-neutral names that do not align with any one gender, among other reasons. Relying solely on binary gender categories without recognizing gender-neutral names can reduce the inclusiveness of gender prediction tasks. We introduce an additional gender category, i.e., "neutral", to study and address potential gender biases in Large Language Models (LLMs).

For Other Balanced SSA datasets, please visit US SSA, Canada SSA, and France SSA.

Dynamic Gender Label Dataset

We observed that each balanced SSA dataset included first names labeled with different genders over the years, as shown in the below table. For example, Victory was recorded as a male name in 1933, a female name in 2000, and as a gender-neutral name in 2016. To further analyze the gender prediction performance of LLMs on first names with varying gender labels over time, we created a dynamic gender label dataset for each country. We selected first names with dynamic gender labels (i.e. names for which the gender association changes over time) from the test set of each balanced SSA dataset.

Dataset Statistics

Please see below and the paper for more details of our curated datasets:

Citation

Please cite the below paper if you intent to use our data for your research:

@inproceedings{you-etal-2024-beyond,
    title = "Beyond Binary Gender Labels: Revealing Gender Bias in {LLM}s through Gender-Neutral Name Predictions",
    author = "You, Zhiwen  and
      Lee, HaeJin  and
      Mishra, Shubhanshu  and
      Jeoung, Sullam  and
      Mishra, Apratim  and
      Kim, Jinseok  and
      Diesner, Jana",
    editor = "Fale{\'n}ska, Agnieszka  and
      Basta, Christine  and
      Costa-juss{\`a}, Marta  and
      Goldfarb-Tarrant, Seraphina  and
      Nozza, Debora",
    booktitle = "Proceedings of the 5th Workshop on Gender Bias in Natural Language Processing (GeBNLP)",
    month = aug,
    year = "2024",
    address = "Bangkok, Thailand",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2024.gebnlp-1.16",
    doi = "10.18653/v1/2024.gebnlp-1.16",
    pages = "255--268",
}

Contact Information

If you have any questions, please email [email protected].