MeetingBank-transcript

MeetingBank数据集源自美国六个主要城市的市议会记录，旨在为现有的数据集提供补充。它包含1366次会议的记录，视频总时长超过3579小时，同时提供会议记录的文本稿、PDF文件（包含会议纪要、议程和其他元数据）。平均每次市议会会议时长为2.6小时，其文本稿包含超过28k个token。该数据集包含6892个段落级别的摘要实例，可用于训练和评估性能，适用于会议内容总结任务。数据集以CC-BY-NC-SA 4.0协议授权。该数据集提供标准化数据操作，并提供音频文件、文本稿、JSON文件、六个系统的摘要以及人工标注等资源。The MeetingBank dataset is derived from city council records of six major U.S. cities, aiming to supplement existing datasets. It contains records of 1366 meetings, with a total video duration of over 3579 hours, and provides meeting transcripts and PDF files (including meeting minutes, agendas and other metadata). The average duration of each city council meeting is 2.6 hours, and its transcript contains over 28k tokens. The dataset includes 6892 paragraph-level summarization instances, which can be used for training and evaluating performance, and is suitable for meeting content summarization tasks. The dataset is licensed under CC-BY-NC-SA 4.0. It provides standardized data operations, as well as resources such as audio files, transcripts, JSON files, summaries from six systems, and manual annotations.