yevgeniy03/home-telecom-callcenter-transcripts-viewer

该公共数据集是从`AIxBlock/92k-real-world-call-center-scripts-english`中的一个源档案`home_ervice_inbound&telecom _outbound.zip`转换而来的扁平化查看器副本，目的是为了在Hugging Face Data Studio中以Parquet表格的形式展示内容。数据集包含3,239个转录本和313,864个推断的轮次，每个转录本和轮次都附带了丰富的元数据，如对话文本、轮次计数、角色推断状态等。数据集还进行了V2清理，添加了可读的`dialogue_text_inferred`列，并通过Amazon Bedrock GPT-OSS-120B进行了角色标注。数据集的语言为英语，大小在1K到10K之间，标签包括呼叫中心、转录本等。

This public dataset is a flattened viewer copy of one source archive from `AIxBlock/92k-real-world-call-center-scripts-english`: `home_ervice_inbound&telecom _outbound.zip`. It was converted so Hugging Face Data Studio can display the contents as regular Parquet tables. The dataset contains 3,239 transcripts and 313,864 inferred turns, each accompanied by rich metadata such as dialogue text, turn counts, role inference status, etc. The dataset also underwent V2 cleanup, adding a readable `dialogue_text_inferred` column and role labeling via Amazon Bedrock GPT-OSS-120B. The dataset is in English, with a size between 1K and 10K, and tags include call-center, transcripts, etc.