Conversational AI Training: Customer Service Intent Labeling
The scope of this project involved preparing high-quality text data for an E-commerce/Customer Service AI model using the Appen platform. Key tasks included Entity (NER) Classification to precisely tag product names, order IDs, and dates within chat transcripts. I also performed Question Answering tasks, extracting specific answers from knowledge base articles to train a robust chatbot Q&A function. Furthermore, the project required Geocoding of address and location data found in customer requests, and general Data Collection for model expansion. I processed an estimated 10,000+ text entries and maintained a consistent quality audit score of 97%+, ensuring strict adherence to complex, multi-layered annotation guidelines for high precision in model training.