Three ways in, depending on what you need: the open pages and files, the explorer, the API. The whole base is reserved for research, with a key.
Video metadata, counts per sentence and per theme, mentions of public figures, records and syntheses, audiences, removals, the list of channels and their orientation.
week.json, trends.json, channels.json, deletions.json, framings.json, political-monthly.json, site-facts.json, video/<id>.json, channel/<slug>.json.Browse videos, transcripts and comments; filter by channel, date or classifier; search by words or by meaning; export according to the key’s tier.
The same tables, over REST, with the same key; the R (youpoldata) and Python (youpol-client) clients cover the common uses.
library(youpoldata)
cli <- youpol_client() # reads YOUPOL_TOKEN
youpol_videos(cli, country = "QC", limit = 10)For every video of the tracked channels, the tables link through the video identifier. Figures for the whole corpus, updated every hour.
Orientation, country, language, platform; closed channels kept.
Title, length, date, channel; reason and date of a removal; timestamped audience measurements.
One turn per anonymous speaker, with the text and the timestamp.
Political label, nine themes with their probability, public figures of the closed list.
Date, likes, replies, author pseudonym; sentences classified as political or not.
Records of the week’s videos, syntheses, records of vanished videos, arguments by theme.
What you get without a key, what needs the explorer or the API.
The channel’s page plots the series and gives its CSV.
no key Go →
The public figure’s record, by month, orientation and channel; the list of videos through the API.
no keyAPI Go →
The explorer filters a video by classifier; the API returns the sentences with their probabilities (Analysis 2 tier).
explorerAPI Go →
The counts are on the video’s record; the text of the comments needs the Research tier.
no keyAPI Go →
CSV or JSON export through the explorer or the R client, at the Research tier.
explorerAPI Go →
The vanished-videos page and its JSON, without a key.
no key Go →
The explorer and the API require a key, issued by the team with an access tier according to the project. Default limits can be adjusted.
Videos, channels, metadata history, aggregates, automatic records. Search in titles.
The previous tier, plus the structure of comments, transcripts and sentences (identifiers, dates, classifier scores), with no text.
The previous tier, plus the text of transcripts, speaker turns and sentences. Search in transcripts, by words or by meaning.
All the data, including the text of comments and their authors’ pseudonyms, corpus exports and embeddings.
Videos stay on their platform and are never redistributed. The text of transcripts and comments is reserved for research, under fair dealing for research (s. 29 of the Copyright Act): never redistributed as such, never for commercial purposes. Records naming a private person are not published. Any use of the data cites the project paper.