Sharmendra/customer-chat-customized

一个包含2000条客户支持聊天对话的合成数据集，涵盖消费电子和技术产品。所有对话均以非正式、口语化的英语自然书写。数据集包括26种产品和25种问题类别，支持多种渠道（如实时聊天、电子邮件聊天、应用内聊天、WhatsApp和网页聊天）。数据集结构包括聊天ID、日期、渠道、客户ID、客户姓名、客服姓名、产品、订单ID、问题类别、情感、解决方案、持续时间、客户满意度评分、完整对话和AI生成的摘要。适用于情感分类、摘要模型微调、意图检测、问题分类、解决方案预测、客户满意度预测和对话研究等用途。

A synthetic dataset of 2,000 customer support chat conversations between customers and agents, covering consumer electronics and technology products. All conversations are naturally written in informal, conversational English. The dataset includes 26 products and 25 issue categories, with support for various channels (e.g., Live Chat, Email Chat, In-App Chat, WhatsApp, Web Chat). The dataset structure includes chat_id, date, channel, customer_id, customer_name, agent_name, product, order_id, issue_category, sentiment, resolution, duration_minutes, csat_score, conversation, and ai_summary. Suitable for training sentiment classifiers, summarisation model fine-tuning, intent detection & issue classification, resolution prediction, customer satisfaction (CSAT) prediction, and dialogue research.