Multilingual Text Annotation for NLP Model Training (Project)
On a self-directed NLP annotation project, I labeled 15,000+ multilingual text samples for sentiment polarity, intent detection, and entity tagging. I applied unified labeling schemas to ensure cross-language consistency. My work improved model performance and supported annotator onboarding with detailed documentation. • Used Label Studio and custom annotation tools for labeling. • Covered English and Yoruba datasets for model training. • Enhanced model F1 scores by ~12%. • Maintained an annotation log for ambiguous labels and edge cases.