archive de youtube sur le lancement d'alerte

Description of the corpus

The corpus describes videos about whistleblowing on the Youtube social media.
Goal of the corpus is the detect automatically new videos (persons or organizations) emitting whistleblowing.
The corpus aims at finding patterns for that purpose.

size of video corpus : 347,544
size of video sample : 22 (one for each topic class)

metadata :

_id : video id
title : video title
channelid : channel id
channeltitle : channel title
datepub : publication date of the video
description : description field of the video
tags : list of keywords for a video
kind : a type , for instance youtube#video
defaultaudiolang : default language of a video
viewcount : number of views for a video
likecount : number of users who likes a video
dislikecount : number of users who do not like a video
commentscount : number of comments
comments : a list of comments :
author : author of a comment
like : number of users who like the comment
message : content of a comment
transcription : video transcription in free text