Hateful Meme Detection Dataset
Abstract
This data list contains two real-world hateful meme datasets: Harm-P and PrideMM.Harm-P includes a large number of memes that satirize, criticize, or attack the U.S. presidential election, political parties, candidates, and policies.The PrideMM dataset was collected by searching and extracting relevant memes from three popular social media platforms—Facebook, Twitter, and Reddit—using hashtags related to LGBTQ+ discussions to filter the content.
Fields of Study & Subjects
Hate Speech and Cyberbullying Detection
Misinformation and Its Impacts
Humor Studies and Applications
Publication details
- Journal / source
- ScienceDB
- Publication date
- 14 January 2026
- Language
- EN
- Document type
- Dataset