The Chinese Courtroom Video Database

As a part of the effort to promote e-governance and judicial transparency, China has been promoting mass online digitalization of court, including an archive of judgment text (司法文书), a platform of online trial videos and live broadcasting (庭审直播), and a judgment implementation tracker (判决执行). China has become one of the few countries that allow cameras in the courtroom. Though a growing number of studies use court decision data, little research has been conducted on the court trial videos. The goal of the Chinese Courtroom Video Database is to meet the needs of those interested in broad research of government policy diffusion, judicial transparency, and judicial behavior in the China context by filling the vacancy in judicial data and providing a new perspective to the existing scholarship. The database includes two sets of data. The first dataset is a catalog of half-million entries of criminal and administrative trial videos in all 31 provinces from January 2013 to February 2019. Each profile records the basic information of a trial video, such as case identification number, date and time of the trial, participants, reason of trial, location of the court, number of views of the video, and other descriptions. The second dataset is a collection of 1,491 audio files of online criminal trials in Yunnan, China. Each audio was downloaded and converted from the original video. The datasets were collected using the Selenium package of Python and Downie, an online stream downloader.

为推进电子政务（e-governance）与司法透明（judicial transparency）建设，中国持续推动法院系统开展大规模在线数字化转型，涵盖司法文书（judgment text）档案库、庭审直播及在线庭审视频平台，以及判决执行追踪系统（judgment implementation tracker）。中国是全球少数允许法庭内架设摄像机的国家之一。
尽管当前已有越来越多的研究依托法院裁判数据展开，但针对庭审视频的系统性研究仍相对匮乏。中国庭审视频数据库（Chinese Courtroom Video Database）旨在填补司法数据资源的空白，为现有学术研究提供崭新视角，以满足学界与实务界针对中国语境下政策扩散、司法透明及司法行为开展全方位研究的需求。
该数据库包含两类数据集：第一类数据集为2013年1月至2019年2月期间全国31个省份的50余万条刑事及行政庭审视频目录，每条目录条目均记录了庭审视频的基础信息，包括案件编号、庭审日期与时段、参与人员、案由、法院所在地、视频播放量及其他相关说明。
第二类数据集为中国云南省1491条在线刑事庭审音频文件，所有音频均由原始庭审视频下载并转换得到。本数据集通过Python的Selenium库（Selenium）及在线流媒体下载工具（Downie）完成采集。