Hateful Meme Detection Dataset

Read full text
Authors: 黄有为
Cited by: 0

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