Tagline: Write Safer Online.
ShieldHER is a real-time online misogyny detection simulator developed for the GNEC Hackathon 2025 Spring. It provides instant visual feedback and suggestions as users type, promoting mindful language and contributing to UN SDGs 5 (Gender Equality) & 10 (Reduced Inequalities).
See ShieldHER in action: (https://www.youtube.com/watch?v=ubyghoNxGHc)
- Real-time (debounced) misogyny detection.
- Visual flagging, likelihood score, and severity.
- Actionable suggestions for harmful text.
- Integrated statistics on the impact of online misogyny.
- Frontend: Next.js (App Router), React, TypeScript, Tailwind CSS
- Backend: Python, Flask
- ML Model:
annahaz/xlm-roberta-base-misogyny-sexism-indomain-mix-bal
- Clone:
git clone https://github.com/YOUR_GITHUB_USERNAME/shieldher.git && cd shieldher - Backend:
activate Python venv,
pip install -r requirements.txtpython app.py - Frontend:
cd ../frontendnpm installnpm run dev - Visit: Go to
http://localhost:3000in your browser (ensure backend is running first).
We envision ShieldHER becoming a cross-platform Browser Extension for seamless, everyday use, and expanding analysis to include text from URLs and Images.