metadata

language:
  - bn
license: cc-by-nc-sa-4.0
task_categories:
  - text-classification
tags:
  - women-safety
  - gender-bias
  - social-media
  - Bangla-NLP
  - sentiment-analysis
pretty_name: 'WoNBias-Partial:  A Balanced Subset of the Original WoNBias Dataset for Review'

WoNBias-Partial: A Balanced Subset of the Original WoNBias Dataset for Review

Dataset Details

Overview

A manually annotated corpus of Bangla designed to benchmark and mitigate gender bias specifically targeting women. The dataset supports research in ethical NLP, hate speech detection, and computational social science.

Dataset Description

Basic Info

Purpose: Detect gender bias against women in Bengali text
Language: Bengali (Bangla)
Labels:
- 0: Neutral (no bias)
- 1: Positive (supportive towards women in general)
- 2: Negative (contains bias against women)

Data Sources

Collected from:

Social media (TikTok, Facebook, Twitter, Instagram)
Online forums
News articles and comments
Government reports

Collection Methods

Semi-automated tools with ethical safeguards:
- Only public content collected
- All personal info removed
- Rate-limited scraping

Partnerships

Developed with help from:

Social media data
Community reachout
Local annotators

Dataset Structure

Format: CSV
Columns:
- Data: Raw Bangla text.
- Label: Integer (0, 1, or 2).

Statistics

Total Samples: 11,178
Label Distribution:
- Neutral (0): 3,644 samples (32.6%)
- Positive (1): 3,506 samples (31.4%)
- Negative (2): 4,028 samples (36.0%)

Ethical Considerations

Data Collection

Anonymization:
- All user identifiers (usernames, URLs, locations) removed
- Metadata sanitized to prevent re-identification
Consent:
- For survey/interview data: Explicit participant consent obtained
- Public posts used under platform terms of service

Content Safeguards

Mental Health Protections:
- Annotators provided with:
  - Psychological support resources
  - Regular breaks when handling toxic content
Trigger Warnings:
- Dataset documentation highlights potentially harmful content

Limitations

Scope Limitations:
- Focuses only on gender bias against women
- Doesn't cover third gender or intersectional biases
Detection Challenges:
- Difficulty identifying implicit bias and contextual nuances
- Performance varies across regional dialects
Language Coverage:
- Currently Bengali-only (may not generalize to other South Asian languages)

Usage

How to Load

from datasets import load_dataset
dataset = load_dataset("won-bias/wonbias-partial-dataset")

Citation

@dataset{wonbias_2024,
  title = {{Wonbias-Partial}: WoNBias-partial: A balanced subset of the original for review},
  author = {Nishat Tafannum, MD. Raisul Islam Aupi},
  year = {2025},
  publisher = {Hugging Face},
  url = {https://huggingface.co/datasets/won-bias/wonbias-partial-dataset},
  version = {1.0},
  license = {CC-BY-NC-SA-4.0},
}

License

This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.

You are free to:

Share — copy and redistribute the material in any medium or format
Adapt — remix, transform, and build upon the material

Under the following terms:

Attribution — You must give appropriate credit, provide a link to the license, and indicate if changes were made.
NonCommercial — You may not use the material for commercial purposes.
ShareAlike — If you remix, transform, or build upon the material, you must distribute your contributions under the same license as the original.