A curated, multi-source dataset for training and evaluating e-commerce conversational AI systems. It covers a broad range of customer intents — from product discovery and order management to returns, tool-augmented responses, and RAG-grounded Q&A — across 16+ product domains.
The train/test split uses prompt-group-level stratified sampling on source ×… See the full description on the dataset page:
https://huggingface.co/datasets/rescommons/Full-Ecom-Chatbot-Dataset.
