Customer Feedback Sentiment and Topic Annotation
Labeled a dataset of 150 customer feedback entries for sentiment and topic classification. Defined clear labeling guidelines with three sentiment classes: positive, negative, neutral. Created five topic categories: billing, service quality, delivery, product quality, general inquiry. Applied consistent rules across all records and documented edge cases. Performed annotation in Excel. Each record included raw text, sentiment label, and topic label. Conducted quality checks by re-labeling a 20 percent sample and comparing results. Achieved 95 percent consistency between initial and review labels. Cleaned text data before labeling by removing duplicates and standardizing format. Maintained a clear audit trail with version control of the dataset. Ensured accuracy, consistency, and adherence to defined guidelines throughout the task.