Memes in the Wild: Assessing the Generalizability of the Hateful Memes Challenge Dataset

Hannah Kirk, Yennie Jun, Paulius Rauba, Gal Wachtel, Ruining Li, Xingjian Bai, Noah Broestl, Martin Doff-Sotta, Aleksandar Shtedritski, Yuki M Asano


Abstract
Hateful memes pose a unique challenge for current machine learning systems because their message is derived from both text- and visual-modalities. To this effect, Facebook released the Hateful Memes Challenge, a dataset of memes with pre-extracted text captions, but it is unclear whether these synthetic examples generalize to ‘memes in the wild’. In this paper, we collect hateful and non-hateful memes from Pinterest to evaluate out-of-sample performance on models pre-trained on the Facebook dataset. We find that ‘memes in the wild’ differ in two key aspects: 1) Captions must be extracted via OCR, injecting noise and diminishing performance of multimodal models, and 2) Memes are more diverse than ‘traditional memes’, including screenshots of conversations or text on a plain background. This paper thus serves as a reality-check for the current benchmark of hateful meme detection and its applicability for detecting real world hate.
Anthology ID:
2021.woah-1.4
Volume:
Proceedings of the 5th Workshop on Online Abuse and Harms (WOAH 2021)
Month:
August
Year:
2021
Address:
Online
Venues:
ACL | IJCNLP | WOAH
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
26–35
Language:
URL:
https://aclanthology.org/2021.woah-1.4
DOI:
10.18653/v1/2021.woah-1.4
Bibkey:
Copy Citation:
PDF:
https://aclanthology.org/2021.woah-1.4.pdf