Twitter History and Image Sharing (THIS) v1.0: aggregate counts derived from a corpus of 14.596 billion geolocated tweets streamed from 1 September 2013 to 14 March 2023, released without tweets or user information. Four tables: tweets and users per country and day (792,313 rows, 250 country codes, 2013-09-01 to 2023-03-14), tweets per country, language and day (15,407,849 rows, 79 languages), new accounts per country and day (943,361 rows, 2006-03-21 to 2023-03-14) and image sharing and availability per country and day (785,061 rows). By Zachary Steinert-Threlkeld (UCLA) and Hanfeng Chen, from the Harvard Dataverse replication data of the Journal of Quantitative Description paper.
Statistics of 2,139,109 YouTube users from the user-information crawl of the re-collection of 3 September 2008 (file 080903user) by Xu Cheng, Cameron Dale and Jiangchuan Liu (Simon Fraser University): for each user name the number of uploaded videos and friends, read from the YouTube API. The file is a list of users with counts; it holds no links between users or videos.
Statistics of 1,062,324 YouTube users from the user-information crawl of the update of 28 to 29 May 2007 (start 28 May 20:48:57, finish 29 May 15:48:54 per the log, no time zone stated) by Xu Cheng, Cameron Dale and Jiangchuan Liu (Simon Fraser University): for each user name the number of uploaded videos, watches and friends, read from the YouTube API. The file is a list of users with counts; it holds no links between users or videos.
1,800 Chinese-language comments from Bilibili videos about population aging, pensions and low fertility. Each comment has its like count, the video it came from, in-vivo glosses in Chinese and English, the codes from two coders, an agreed final code and a process-model stage. The package also includes metadata for the source videos and the codebook. Zhangwei Zheng (Sunway University) deposited it on Harvard Dataverse under CC0 as replication data for a study of Bilibili comment discourse on aging.
Statistics of the top 1,000 Twitch channels over one year: watch time, stream time, peak and average viewers, followers, followers gained, views gained, whether the channel is a partner (978 are) or marked mature (230 are) and its stream language (21 languages; most frequent English (485), Korean (77), Russian (74), Spanish (68)). Compiled by Aayush Mishra and published on Kaggle as "Top Streamers on Twitch". One row per channel; the file carries no individual viewers and no dates.
EDTT (Eating Disorder TikTok) public dataset: 85 pro-eating-disorder TikTok posts (video (47), photos (38)), each described by a written summary and coded by subject-matter experts for eating-disorder features (Eating Disorder Examination subscale and diagnostic items), visual and audio elements and audio measures (tempo, valence, arousal, harmonic ratio, mode). Posts were found by text search and by recommendation-based surfacing. The file carries no post IDs, links, user names or dates.
Twitter (X)
YouTube
Twitch
TikTok