Summary

This dataset contains TikTok comments and replies identified as antisemitic. This dataset contains metadata about each comment, such as user profile, engagement metrics, and geographical data, to support comprehensive analysis.

Date Fields

The dataset is structured as a CSV file with the following columns:

post_id: Unique numerical identifier for the TikTok post comment_id: Unique numerical identifier for the comment or reply parent_id: Reference to the comment_id of the comment that this comment is replying to

username: Unique identifier for the user who published the comment or reply nickname: Display name of the user who published the comment or reply bio: Bio or description of the user who published the comment or reply avatar: URL that points to the avatar associated with the user's profile

text: Full text of the comment or reply

region: Geographical region where the user published the comment or reply

likes: Number of likes that the comment or reply received reply_count: Number of replies that the comment or reply generated liked_by_creator: Boolean indicating whether the post author "liked" the comment or reply

created: Timestamp when the comment or reply was published

Dataset Creation

Data Collection: data collection via the TikTok API, focusing on post interactions
Keyword Filtering: User interactions such as comments and replies, and user profile information such as nicknames and descriptions were filtered using three distinct keywords groups:
- Israel-related symbols: Emojis such as the Flag of Israel (🇮🇱) and the Star of David (✡️) were used to identify relevant content and possible misuses
- Yellow Ribbon: The yellow ribbon emoji (🎗️), a symbol of hope and support for those who are held in captivity, was also included to identify relevant content
- Antisemetic Keywords: Specific keywords and phrases commonly associated with antisemitism were used to filter and identify content with antisemetic undertones
Data scraping: Comments, replies, and user profile information identified to contain antisemetic content were subjected to targeted data scraping to collect detailed information, including the whole thread history and user profile metadata.

Limitations

This dataset only includes text-based content from user interactions and user profiles. This dataset does not include photo or video content.