Kinyarwanda Hate Speech Detection — Sentiment Annotation
Labeled thousands of Kinyarwanda-language text samples with sentiment and toxicity tags to train a hate speech detection classifier. Paid close attention to cultural and linguistic nuance while adhering to annotation style guides and consistency requirements. Delivered large, high-quality labeled data batches for model training evaluation. • Focused on an underrepresented African language (Kinyarwanda) for NLP. • Emphasized subjective content labeling for hate speech detection. • Maintained accuracy through self-QA and batch-level review. • Supported supervised learning project workflows for sentiment annotation.