Multilingual Counter Speech
An AI-driven system that detects and analyses hate speech across multiple languages and generates culturally relevant counter-narratives to promote positive discourse.
FIG. 01 PipelineNA · 2026
Answering hate in its own language
The idea
A useful reply has to fit the language and the community it is aimed at.
The system detects hateful content, works out the language and the group being targeted, and generates a culturally relevant counter-narrative. An interactive Gradio interface runs the whole pipeline on real text.
- Hate speech detection across languages
- Automatic language and target-group classification
- Sentiment and toxicity analysis
- TF-IDF vectorization and cosine similarity
- An interactive Gradio interface
- Type
- Research
- Published
- IJIRCCE, Dec 2024
- Stack
- Python, NLP, Gradio
- Scope
- Multilingual
How it works
Detect, classify, respond
Detect
Sentiment and toxicity analysis help flag hateful content.
Classify
The language and the targeted group are classified automatically.
Respond
A culturally relevant counter-narrative is generated for that context.
Try it
An interactive Gradio interface runs the pipeline on real text.