Hate Speech and Toxic Comment Detector

MSc Data Science | CMP-L016 Deep Learning Applications | University of Roehampton

This tool uses HateBERT fine-tuned on the Jigsaw Toxic Comment dataset to classify text as toxic or non-toxic. Token-level SHAP explanations show which words most influenced the prediction.

Input Text

Example inputs - click to load

Classification Result

Confidence Scores

SHAP Token Importance (red = pushes toward Toxic, green = toward Non-Toxic)

Research Disclaimer: This tool is for academic research purposes only. It does not constitute a content moderation decision. False positives and false negatives are possible.