TikTok Bots and Clickworker

Published: 15-10-2020| Version 1 | DOI: 10.17632/gtnbx2h84c.1
Fabian Schenk


A TikTok dataset of 10131 partially labeled data with 1200 negative, 1788 positive and 7.143 unlabeled records. Each with information about their last 20 uploaded videos if available. The limitation to the last 20 videos had to be done because otherwise the data colletion process would have consumed to much time with the technique I have used. Unfortunately names of accounts and other textual descriptions such as account and video descriptions are not available due to data protection concerns. (Yes I'm from Germany...) The data was collected by my own C# Application which used Appium on a smartphone to automate the data gathering process. How colletion and labeling was done: - Unlabeled: Collected from 2 different accounts by iterating over their followerlists - Labeled as Bots and Clickworkers: Accounts were bought on 6 different internet pages for a TikTok account with no other follwers or contents. - Labeled as Human/No Bot or Clickworker: 1. Friends or friends of friends. 2. Accounts who wrote contextual comments. 3. Accounts who recommended a video to other accounts in comments. 4. The Accounts who were linked in comments. 5. Accounts of publicly known personalities or insitutions. 6. Accounts whose names could be read in Videos which were exported from TikTok and uploaded to Instagram with "TikTok"-related Hashtags Additionally to the available Features I've already created some averages of views, comments, likes or forwardings for the videos of each collected account. To view in Excel make sure you select Tabstop (\t) as delimiter and import every column as "TEXT" (Use shift to select all columns at once). Otherwise Excel will try to interpret some numbers as dates and ruin the data. The videos can be assigned to the accounts using the primare key column "Id" of the accounts and the foreign key column "AccountId" of the videos. If you have any questions please feel free to contact me.