diner-call
dinercall-ner
dinercall‑ner
A synthetic bilingual (Chinese / English) dataset for restaurant‑reservation NER on ASR‑like utterances
📖 Overview
dinercall-ner contains 20 000 automatically generated reservation requests that mimic spoken input from an Automatic Speech Recognition (ASR) system.
Languages: 70 % Mandarin (Traditional Chinese), 30 % English.
File format: dataset.parquet (columnar, ready for datasets or pandas).
Each row provides the raw (error‑prone) utterance and… See the full description on the dataset page: https://huggingface.co/datasets/Luigi/dinercall-ner.dinercall-intent
DinerCall Intent
DinerCall Intent is a fully synthesized intent classification dataset composed of answering machine messages from a diner restaurant. The transcriptions simulate ASR outputs (with typical speech-to-text errors). Over 80% of the messages are in Taiwan Chinese, with the remainder in English. The dataset was generated using OpenAI ChatGPT.
Format: CSVColumns:
text: ASR-style transcription of the diner’s answering machine message.
intent: The call’s intent… See the full description on the dataset page: https://huggingface.co/datasets/Luigi/dinercall-intent.