Chatbot Datasets for Low-Resource E-Commerce - IEEE TSC 2025

This dataset collection supports the paper \Low-Resource Domain Adaptation for Chatbots: Enabling Scalable AI in E-Commerce Environments\, submitted to IEEE Transactions on Services Computing (TSC). It includes three datasets:MultiRetail-Mix (synthetic, multilingual)Retail-Chat (real, English)E-ComAssist (synthetic, multilingual, multi-intent)These datasets span diverse retail scenarios such as product inquiries, returns, delivery issues, and pricing queries across multiple languages (English, Spanish, and Hindi). They are designed to benchmark and train chatbot models under low-resource and domain-adaptive conditions. Ideal for tasks such as intent classification, multilingual NLU, and chatbot evaluation in service computing and AI for retail.All datasets are licensed under CC BY 4.0 and structured in CSV format with fields like utterance, intent, language, domain, and source_type.