@inproceedings{toraman-etal-2022-large,
title = "Large-Scale Hate Speech Detection with Cross-Domain Transfer",
author = "Toraman, Cagri and
{\c{S}}ahinu{\c{c}}, Furkan and
Yilmaz, Eyup",
booktitle = "Proceedings of the Thirteenth Language Resources and Evaluation Conference",
month = jun,
year = "2022",
address = "Marseille, France",
publisher = "European Language Resources Association",
url = "https://aclanthology.org/2022.lrec-1.238",
pages = "2215--2225",
abstract = "The performance of hate speech detection models relies on the datasets on which the models are trained. Existing datasets are mostly prepared with a limited number of instances or hate domains that define hate topics. This hinders large-scale analysis and transfer learning with respect to hate domains. In this study, we construct large-scale tweet datasets for hate speech detection in English and a low-resource language, Turkish, consisting of human-labeled 100k tweets per each. Our datasets are designed to have equal number of tweets distributed over five domains. The experimental results supported by statistical tests show that Transformer-based language models outperform conventional bag-of-words and neural models by at least 5{\%} in English and 10{\%} in Turkish for large-scale hate speech detection. The performance is also scalable to different training sizes, such that 98{\%} of performance in English, and 97{\%} in Turkish, are recovered when 20{\%} of training instances are used. We further examine the generalization ability of cross-domain transfer among hate domains. We show that 96{\%} of the performance of a target domain in average is recovered by other domains for English, and 92{\%} for Turkish. Gender and religion are more successful to generalize to other domains, while sports fail most.",
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="toraman-etal-2022-large">
<titleInfo>
<title>Large-Scale Hate Speech Detection with Cross-Domain Transfer</title>
</titleInfo>
<name type="personal">
<namePart type="given">Cagri</namePart>
<namePart type="family">Toraman</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Furkan</namePart>
<namePart type="family">Şahinuç</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Eyup</namePart>
<namePart type="family">Yilmaz</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2022-06</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the Thirteenth Language Resources and Evaluation Conference</title>
</titleInfo>
<originInfo>
<publisher>European Language Resources Association</publisher>
<place>
<placeTerm type="text">Marseille, France</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>The performance of hate speech detection models relies on the datasets on which the models are trained. Existing datasets are mostly prepared with a limited number of instances or hate domains that define hate topics. This hinders large-scale analysis and transfer learning with respect to hate domains. In this study, we construct large-scale tweet datasets for hate speech detection in English and a low-resource language, Turkish, consisting of human-labeled 100k tweets per each. Our datasets are designed to have equal number of tweets distributed over five domains. The experimental results supported by statistical tests show that Transformer-based language models outperform conventional bag-of-words and neural models by at least 5% in English and 10% in Turkish for large-scale hate speech detection. The performance is also scalable to different training sizes, such that 98% of performance in English, and 97% in Turkish, are recovered when 20% of training instances are used. We further examine the generalization ability of cross-domain transfer among hate domains. We show that 96% of the performance of a target domain in average is recovered by other domains for English, and 92% for Turkish. Gender and religion are more successful to generalize to other domains, while sports fail most.</abstract>
<identifier type="citekey">toraman-etal-2022-large</identifier>
<location>
<url>https://aclanthology.org/2022.lrec-1.238</url>
</location>
<part>
<date>2022-06</date>
<extent unit="page">
<start>2215</start>
<end>2225</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Large-Scale Hate Speech Detection with Cross-Domain Transfer
%A Toraman, Cagri
%A Şahinuç, Furkan
%A Yilmaz, Eyup
%S Proceedings of the Thirteenth Language Resources and Evaluation Conference
%D 2022
%8 June
%I European Language Resources Association
%C Marseille, France
%F toraman-etal-2022-large
%X The performance of hate speech detection models relies on the datasets on which the models are trained. Existing datasets are mostly prepared with a limited number of instances or hate domains that define hate topics. This hinders large-scale analysis and transfer learning with respect to hate domains. In this study, we construct large-scale tweet datasets for hate speech detection in English and a low-resource language, Turkish, consisting of human-labeled 100k tweets per each. Our datasets are designed to have equal number of tweets distributed over five domains. The experimental results supported by statistical tests show that Transformer-based language models outperform conventional bag-of-words and neural models by at least 5% in English and 10% in Turkish for large-scale hate speech detection. The performance is also scalable to different training sizes, such that 98% of performance in English, and 97% in Turkish, are recovered when 20% of training instances are used. We further examine the generalization ability of cross-domain transfer among hate domains. We show that 96% of the performance of a target domain in average is recovered by other domains for English, and 92% for Turkish. Gender and religion are more successful to generalize to other domains, while sports fail most.
%U https://aclanthology.org/2022.lrec-1.238
%P 2215-2225
Markdown (Informal)
[Large-Scale Hate Speech Detection with Cross-Domain Transfer](https://aclanthology.org/2022.lrec-1.238) (Toraman et al., LREC 2022)
ACL