Text Sentiment & Category Labeling Dataset
TextClassification
Annotated 500 user-generated product reviews with two independent label types: a three-class sentiment label (positive, neutral, negative) and one of eight topic category labels (electronics, clothing, home & kitchen, etc.). Applied structured annotation guidelines to handle edge cases including sarcasm, mixed sentiment, and ambiguous categories. Performed a 10% re-labeling QA pass achieving ~94% inter-annotator agreement. Delivered a structured CSV dataset with confidence scores and quality flags, simulating real-world text classification workflows.
Annotated 500 user-generated product reviews with two independent label types: a three-class sentiment label (positive, neutral, negative) and one of eight topic category labels (electronics, clothing, home & kitchen, etc.). Applied structured annotation guidelines to handle edge cases including sarcasm, mixed sentiment, and ambiguous categories. Performed a 10% re-labeling QA pass achieving ~94% inter-annotator agreement. Delivered a structured CSV dataset with confidence scores and quality flags, simulating real-world text classification workflows.
2025 - Present