@inproceedings{aggarwal-etal-2024-text,
title = "Text or Image? What is More Important in Cross-Domain Generalization Capabilities of Hate Meme Detection Models?",
author = {Aggarwal, Piush and
Mehrabanian, Jawar and
Huang, Weigang and
Alacam, {\"O}zge and
Zesch, Torsten},
editor = "Graham, Yvette and
Purver, Matthew",
booktitle = "Findings of the Association for Computational Linguistics: EACL 2024",
month = mar,
year = "2024",
address = "St. Julian{'}s, Malta",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-eacl.8",
pages = "104--117",
abstract = "This paper delves into the formidable challenge of cross-domain generalization in multimodal hate meme detection, presenting compelling findings. We provide evidence supporting the hypothesis that only the textual component of hateful memes enables the multimodal classifier to generalize across different domains, while the image component proves highly sensitive to a specific training dataset. The evidence includes demonstrations showing that hate-text classifiers perform similarly to hate-meme classifiers in a zero-shot setting. Simultaneously, the introduction of captions generated from images of memes to the hate-meme classifier worsens performance by an average F1 of 0.02. Through blackbox explanations, we identify a substantial contribution of the text modality (average of 83{\%}), which diminishes with the introduction of meme{'}s image captions (52{\%}). Additionally, our evaluation on a newly created confounder dataset reveals higher performance on text confounders as compared to image confounders with average ∆F1 of 0.18.",
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="aggarwal-etal-2024-text">
<titleInfo>
<title>Text or Image? What is More Important in Cross-Domain Generalization Capabilities of Hate Meme Detection Models?</title>
</titleInfo>
<name type="personal">
<namePart type="given">Piush</namePart>
<namePart type="family">Aggarwal</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jawar</namePart>
<namePart type="family">Mehrabanian</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Weigang</namePart>
<namePart type="family">Huang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Özge</namePart>
<namePart type="family">Alacam</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Torsten</namePart>
<namePart type="family">Zesch</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2024-03</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Findings of the Association for Computational Linguistics: EACL 2024</title>
</titleInfo>
<name type="personal">
<namePart type="given">Yvette</namePart>
<namePart type="family">Graham</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Matthew</namePart>
<namePart type="family">Purver</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">St. Julian’s, Malta</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>This paper delves into the formidable challenge of cross-domain generalization in multimodal hate meme detection, presenting compelling findings. We provide evidence supporting the hypothesis that only the textual component of hateful memes enables the multimodal classifier to generalize across different domains, while the image component proves highly sensitive to a specific training dataset. The evidence includes demonstrations showing that hate-text classifiers perform similarly to hate-meme classifiers in a zero-shot setting. Simultaneously, the introduction of captions generated from images of memes to the hate-meme classifier worsens performance by an average F1 of 0.02. Through blackbox explanations, we identify a substantial contribution of the text modality (average of 83%), which diminishes with the introduction of meme’s image captions (52%). Additionally, our evaluation on a newly created confounder dataset reveals higher performance on text confounders as compared to image confounders with average ∆F1 of 0.18.</abstract>
<identifier type="citekey">aggarwal-etal-2024-text</identifier>
<location>
<url>https://aclanthology.org/2024.findings-eacl.8</url>
</location>
<part>
<date>2024-03</date>
<extent unit="page">
<start>104</start>
<end>117</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Text or Image? What is More Important in Cross-Domain Generalization Capabilities of Hate Meme Detection Models?
%A Aggarwal, Piush
%A Mehrabanian, Jawar
%A Huang, Weigang
%A Alacam, Özge
%A Zesch, Torsten
%Y Graham, Yvette
%Y Purver, Matthew
%S Findings of the Association for Computational Linguistics: EACL 2024
%D 2024
%8 March
%I Association for Computational Linguistics
%C St. Julian’s, Malta
%F aggarwal-etal-2024-text
%X This paper delves into the formidable challenge of cross-domain generalization in multimodal hate meme detection, presenting compelling findings. We provide evidence supporting the hypothesis that only the textual component of hateful memes enables the multimodal classifier to generalize across different domains, while the image component proves highly sensitive to a specific training dataset. The evidence includes demonstrations showing that hate-text classifiers perform similarly to hate-meme classifiers in a zero-shot setting. Simultaneously, the introduction of captions generated from images of memes to the hate-meme classifier worsens performance by an average F1 of 0.02. Through blackbox explanations, we identify a substantial contribution of the text modality (average of 83%), which diminishes with the introduction of meme’s image captions (52%). Additionally, our evaluation on a newly created confounder dataset reveals higher performance on text confounders as compared to image confounders with average ∆F1 of 0.18.
%U https://aclanthology.org/2024.findings-eacl.8
%P 104-117
Markdown (Informal)
[Text or Image? What is More Important in Cross-Domain Generalization Capabilities of Hate Meme Detection Models?](https://aclanthology.org/2024.findings-eacl.8) (Aggarwal et al., Findings 2024)
ACL