How well do hate speech, toxicity, abusive and offensive language classification models generalize across datasets?

A considerable body of research deals with the automatic identification of hate speech and related phenomena. However, cross-dataset model generalization remains a challenge. In this context, we address two still open central questions: (i) to what extent does the generalization depend on the model...

Full description

Bibliographic Details
Authors: Fortuna, Paula, Soler Company, Juan, Wanner, Leo
Format: article
Status:Published version
Publication Date:2021
Country:España
Institution:Varias* (Consorci de Biblioteques Universitáries de Catalunya, Centre de Serveis Científics i Acadèmics de Catalunya)
Repository:Recercat. Dipósit de la Recerca de Catalunya
OAI Identifier:oai:recercat.cat:10230/53157
Online Access:http://hdl.handle.net/10230/53157
http://dx.doi.org/10.1016/j.ipm.2021.102524
Access Level:Open access
Keyword:Hate speech
Offensive language
Classification
Generalization
id ES_14c3c66fc6331f433c8e78b22debd873
oai_identifier_str oai:recercat.cat:10230/53157
network_acronym_str ES
network_name_str España
repository_id_str
spelling How well do hate speech, toxicity, abusive and offensive language classification models generalize across datasets?Fortuna, PaulaSoler Company, JuanWanner, LeoHate speechOffensive languageClassificationGeneralizationA considerable body of research deals with the automatic identification of hate speech and related phenomena. However, cross-dataset model generalization remains a challenge. In this context, we address two still open central questions: (i) to what extent does the generalization depend on the model and the composition and annotation of the training data in terms of different categories?, and (ii) do specific features of the datasets or models influence the generalization potential? To answer (i), we experiment with BERT, ALBERT, fastText, and SVM models trained on nine common public English datasets, whose class (or category) labels are standardized (and thus made comparable), in intra- and cross-dataset setups. The experiments show that indeed the generalization varies from model to model and that some of the categories (e.g., ‘toxic’, ‘abusive’, or ‘offensive’) serve better as cross-dataset training categories than others (e.g., ‘hate speech’). To answer (ii), we use a Random Forest model for assessing the relevance of different model and dataset features during the prediction of the performance of 450 BERT, 450 ALBERT, 450 fastText, and 348 SVM binary abusive language classifiers (1698 in total). We find that in order to generalize well, a model already needs to perform well in an intra-dataset scenario. Furthermore, we find that some other parameters are equally decisive for the success of the generalization, including, e.g., the training and target categories and the percentage of the out-of-domain vocabulary.The first author is supported by the research grant SFRH/BD/143623/2019, provided by the Portuguese national funding agency for science, research and technology, Fundação para a Ciência e a Tecnologia (FCT), within the scope of Operational Program Human Capital (POCH), supported by the European Social Fund and by national funds from MCTES. The work of the second and third authors has been supported by the European Commission in the context of the H2020 Research Program under the contract numbers 700024 and 786731.Elsevier202220222021info:eu-repo/semantics/articleinfo:eu-repo/semantics/publishedVersionapplication/pdfapplication/pdfhttp://hdl.handle.net/10230/53157http://dx.doi.org/10.1016/j.ipm.2021.102524reponame:Recercat. Dipósit de la Recerca de Catalunyainstname:Varias* (Consorci de Biblioteques Universitáries de Catalunya, Centre de Serveis Científics i Acadèmics de Catalunya)InglésInformation Processing and Management. 2021;58(3):102524.info:eu-repo/grantAgreement/EC/H2020/700024info:eu-repo/grantAgreement/EC/H2020/786731© 2021 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND licensehttps://creativecommons.org/licenses/by-nc-nd/4.0/info:eu-repo/semantics/openAccessoai:recercat.cat:10230/531572026-05-29T05:05:01Z
dc.title.none.fl_str_mv How well do hate speech, toxicity, abusive and offensive language classification models generalize across datasets?
title How well do hate speech, toxicity, abusive and offensive language classification models generalize across datasets?
spellingShingle How well do hate speech, toxicity, abusive and offensive language classification models generalize across datasets?
Fortuna, Paula
Hate speech
Offensive language
Classification
Generalization
title_short How well do hate speech, toxicity, abusive and offensive language classification models generalize across datasets?
title_full How well do hate speech, toxicity, abusive and offensive language classification models generalize across datasets?
title_fullStr How well do hate speech, toxicity, abusive and offensive language classification models generalize across datasets?
title_full_unstemmed How well do hate speech, toxicity, abusive and offensive language classification models generalize across datasets?
title_sort How well do hate speech, toxicity, abusive and offensive language classification models generalize across datasets?
dc.creator.none.fl_str_mv Fortuna, Paula
Soler Company, Juan
Wanner, Leo
author Fortuna, Paula
author_facet Fortuna, Paula
Soler Company, Juan
Wanner, Leo
author_role author
author2 Soler Company, Juan
Wanner, Leo
author2_role author
author
dc.subject.none.fl_str_mv Hate speech
Offensive language
Classification
Generalization
topic Hate speech
Offensive language
Classification
Generalization
description A considerable body of research deals with the automatic identification of hate speech and related phenomena. However, cross-dataset model generalization remains a challenge. In this context, we address two still open central questions: (i) to what extent does the generalization depend on the model and the composition and annotation of the training data in terms of different categories?, and (ii) do specific features of the datasets or models influence the generalization potential? To answer (i), we experiment with BERT, ALBERT, fastText, and SVM models trained on nine common public English datasets, whose class (or category) labels are standardized (and thus made comparable), in intra- and cross-dataset setups. The experiments show that indeed the generalization varies from model to model and that some of the categories (e.g., ‘toxic’, ‘abusive’, or ‘offensive’) serve better as cross-dataset training categories than others (e.g., ‘hate speech’). To answer (ii), we use a Random Forest model for assessing the relevance of different model and dataset features during the prediction of the performance of 450 BERT, 450 ALBERT, 450 fastText, and 348 SVM binary abusive language classifiers (1698 in total). We find that in order to generalize well, a model already needs to perform well in an intra-dataset scenario. Furthermore, we find that some other parameters are equally decisive for the success of the generalization, including, e.g., the training and target categories and the percentage of the out-of-domain vocabulary.
publishDate 2021
dc.date.none.fl_str_mv 2021
2022
2022
dc.type.none.fl_str_mv info:eu-repo/semantics/article
info:eu-repo/semantics/publishedVersion
format article
status_str publishedVersion
dc.identifier.none.fl_str_mv http://hdl.handle.net/10230/53157
http://dx.doi.org/10.1016/j.ipm.2021.102524
url http://hdl.handle.net/10230/53157
http://dx.doi.org/10.1016/j.ipm.2021.102524
dc.language.none.fl_str_mv Inglés
language_invalid_str_mv Inglés
dc.relation.none.fl_str_mv Information Processing and Management. 2021;58(3):102524.
info:eu-repo/grantAgreement/EC/H2020/700024
info:eu-repo/grantAgreement/EC/H2020/786731
dc.rights.none.fl_str_mv https://creativecommons.org/licenses/by-nc-nd/4.0/
info:eu-repo/semantics/openAccess
rights_invalid_str_mv https://creativecommons.org/licenses/by-nc-nd/4.0/
eu_rights_str_mv openAccess
dc.format.none.fl_str_mv application/pdf
application/pdf
dc.publisher.none.fl_str_mv Elsevier
publisher.none.fl_str_mv Elsevier
dc.source.none.fl_str_mv reponame:Recercat. Dipósit de la Recerca de Catalunya
instname:Varias* (Consorci de Biblioteques Universitáries de Catalunya, Centre de Serveis Científics i Acadèmics de Catalunya)
instname_str Varias* (Consorci de Biblioteques Universitáries de Catalunya, Centre de Serveis Científics i Acadèmics de Catalunya)
reponame_str Recercat. Dipósit de la Recerca de Catalunya
collection Recercat. Dipósit de la Recerca de Catalunya
repository.name.fl_str_mv
repository.mail.fl_str_mv
_version_ 1869403755185700864
score 15.812429

Similar Items