Dataset Card:客服工單分類與結構化抽取(合成資料)
資料集摘要
繁體中文(台灣)電商客服工單合成資料集,用於六分類意圖分類 + 結構化欄位抽取任務。全部資料由 teacher LLM(Claude / GPT,經 OpenRouter 呼叫)依 schema-first structured output 生成,不含任何真實客服工單,無個資/版權疑慮。共 3,148 筆,切分為 train / validation / test 三份。
ds = load_dataset("albertkingdom/custom-service-ticket-zh-tw", data_files={
"train": "train.jsonl",
"validation": "validation.jsonl",
"test": "test.jsonl",
})… See the full description on the dataset page:
https://huggingface.co/datasets/albertkingdom/custom-service-ticket-zh-tw.