Systematic keyword and bias analyses in hate speech detection

[EN] Hate speech detection refers broadly to the automatic identification of language that may be considered discriminatory against certain groups of people. The goal is to help online platforms to identify and remove harmful content. Humans are usually capable of detecting hatred in critical cases,...

Descripción completa

Detalles Bibliográficos
Autores: Peña-Sarracén, Gretel Liz de la, Rosso, Paolo
Tipo de recurso: artículo
Fecha de publicación:2023
País:España
Institución:Universitat Politècnica de València (UPV)
Repositorio:RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Idioma:inglés
OAI Identifier:oai:riunet.upv.es:10251/204644
Acceso en línea:https://riunet.upv.es/handle/10251/204644
Access Level:acceso abierto
Palabra clave:Hate speech detection
Keyword extraction
Bias analysis
Bias mitigation
LENGUAJES Y SISTEMAS INFORMATICOS
id ES_9cbabf7c1a0c8cc707f2a830e2ead97a
oai_identifier_str oai:riunet.upv.es:10251/204644
network_acronym_str ES
network_name_str España
repository_id_str
spelling Systematic keyword and bias analyses in hate speech detectionPeña-Sarracén, Gretel Liz de laRosso, PaoloHate speech detectionKeyword extractionBias analysisBias mitigationLENGUAJES Y SISTEMAS INFORMATICOS[EN] Hate speech detection refers broadly to the automatic identification of language that may be considered discriminatory against certain groups of people. The goal is to help online platforms to identify and remove harmful content. Humans are usually capable of detecting hatred in critical cases, such as when the hatred is non-explicit, but how do computer models address this situation? In this work, we aim to contribute to the understanding of ethical issues related to hate speech by analysing two transformer-based models trained to detect hate speech. Our study focuses on analysing the relationship between these models and a set of hateful keywords extracted from the three well-known datasets. For the extraction of the keywords, we propose a metric that takes into account the division among classes to favour the most common words in hateful contexts. In our experiments, we first compared the overlap between the extracted keywords with the words to which the models pay the most attention in decision-making. On the other hand, we investigate the bias of the models towards the extracted keywords. For the bias analysis, we characterize and use two metrics and evaluate two strategies to try to mitigate the bias. Surprisingly, we show that over 50% of the salient words of the models are not hateful and that there is a higher number of hateful words among the extracted keywords. However, we show that the models appear to be biased towards the extracted keywords. Experimental results suggest that fitting models with hateful texts that do not contain any of the keywords can reduce bias and improve the performance of the models.This work was done in the framework of the research project on Fairness and Transparency for equitable NLP applications in social media, funded by MCIN/AEI/10.13039/501100011033 and by ERDF, EU A way of making EuropePI.ElsevierDepartamento de Sistemas Informáticos y ComputaciónEscuela Técnica Superior de Ingeniería InformáticaCentro de Investigación Pattern Recognition and Human Language TechnologyAGENCIA ESTATAL DE INVESTIGACIONEuropean Regional Development FundUniversitat Politècnica de ValènciaRepositorio Institucional de la Universitat Politècnica de València Riunet20232023-09-01journal articlehttp://purl.org/coar/resource_type/c_6501VoRhttp://purl.org/coar/version/c_970fb48d4fbd8a85info:eu-repo/semantics/articleapplication/pdfhttps://riunet.upv.es/handle/10251/204644reponame:RiuNet. Repositorio Institucional de la Universitat Politécnica de Valénciainstname:Universitat Politècnica de València (UPV)InglésengAgencia Estatal de Investigación http://dx.doi.org/10.13039/501100011033 Plan Estatal de Investigación Científica y Técnica y de Innovación 2021-2023 PID2021-124361OB-C31 FAIRTRANSNLP-STEREOTYPES: IDENTIFICACION DE ESTEREOTIPOS Y PREJUICIOS Y DESARROLLO DE SISTEMAS EQUITATIVOSopen accesshttp://purl.org/coar/access_right/c_abf2Reconocimiento (by)http://creativecommons.org/licenses/by/4.0/info:eu-repo/semantics/openAccessoai:riunet.upv.es:10251/2046442026-06-13T07:49:27Z
dc.title.none.fl_str_mv Systematic keyword and bias analyses in hate speech detection
title Systematic keyword and bias analyses in hate speech detection
spellingShingle Systematic keyword and bias analyses in hate speech detection
Peña-Sarracén, Gretel Liz de la
Hate speech detection
Keyword extraction
Bias analysis
Bias mitigation
LENGUAJES Y SISTEMAS INFORMATICOS
title_short Systematic keyword and bias analyses in hate speech detection
title_full Systematic keyword and bias analyses in hate speech detection
title_fullStr Systematic keyword and bias analyses in hate speech detection
title_full_unstemmed Systematic keyword and bias analyses in hate speech detection
title_sort Systematic keyword and bias analyses in hate speech detection
dc.creator.none.fl_str_mv Peña-Sarracén, Gretel Liz de la
Rosso, Paolo
author Peña-Sarracén, Gretel Liz de la
author_facet Peña-Sarracén, Gretel Liz de la
Rosso, Paolo
author_role author
author2 Rosso, Paolo
author2_role author
dc.contributor.none.fl_str_mv Departamento de Sistemas Informáticos y Computación
Escuela Técnica Superior de Ingeniería Informática
Centro de Investigación Pattern Recognition and Human Language Technology
AGENCIA ESTATAL DE INVESTIGACION
European Regional Development Fund
Universitat Politècnica de València
Repositorio Institucional de la Universitat Politècnica de València Riunet
dc.subject.none.fl_str_mv Hate speech detection
Keyword extraction
Bias analysis
Bias mitigation
LENGUAJES Y SISTEMAS INFORMATICOS
topic Hate speech detection
Keyword extraction
Bias analysis
Bias mitigation
LENGUAJES Y SISTEMAS INFORMATICOS
description [EN] Hate speech detection refers broadly to the automatic identification of language that may be considered discriminatory against certain groups of people. The goal is to help online platforms to identify and remove harmful content. Humans are usually capable of detecting hatred in critical cases, such as when the hatred is non-explicit, but how do computer models address this situation? In this work, we aim to contribute to the understanding of ethical issues related to hate speech by analysing two transformer-based models trained to detect hate speech. Our study focuses on analysing the relationship between these models and a set of hateful keywords extracted from the three well-known datasets. For the extraction of the keywords, we propose a metric that takes into account the division among classes to favour the most common words in hateful contexts. In our experiments, we first compared the overlap between the extracted keywords with the words to which the models pay the most attention in decision-making. On the other hand, we investigate the bias of the models towards the extracted keywords. For the bias analysis, we characterize and use two metrics and evaluate two strategies to try to mitigate the bias. Surprisingly, we show that over 50% of the salient words of the models are not hateful and that there is a higher number of hateful words among the extracted keywords. However, we show that the models appear to be biased towards the extracted keywords. Experimental results suggest that fitting models with hateful texts that do not contain any of the keywords can reduce bias and improve the performance of the models.
publishDate 2023
dc.date.none.fl_str_mv 2023
2023-09-01
dc.type.none.fl_str_mv journal article
http://purl.org/coar/resource_type/c_6501
VoR
http://purl.org/coar/version/c_970fb48d4fbd8a85
dc.type.openaire.fl_str_mv info:eu-repo/semantics/article
format article
dc.identifier.none.fl_str_mv https://riunet.upv.es/handle/10251/204644
url https://riunet.upv.es/handle/10251/204644
dc.language.none.fl_str_mv Inglés
eng
language_invalid_str_mv Inglés
language eng
dc.relation.none.fl_str_mv Agencia Estatal de Investigación http://dx.doi.org/10.13039/501100011033 Plan Estatal de Investigación Científica y Técnica y de Innovación 2021-2023 PID2021-124361OB-C31 FAIRTRANSNLP-STEREOTYPES: IDENTIFICACION DE ESTEREOTIPOS Y PREJUICIOS Y DESARROLLO DE SISTEMAS EQUITATIVOS
dc.rights.none.fl_str_mv open access
http://purl.org/coar/access_right/c_abf2
Reconocimiento (by)
http://creativecommons.org/licenses/by/4.0/
dc.rights.openaire.fl_str_mv info:eu-repo/semantics/openAccess
rights_invalid_str_mv open access
http://purl.org/coar/access_right/c_abf2
Reconocimiento (by)
http://creativecommons.org/licenses/by/4.0/
eu_rights_str_mv openAccess
dc.format.none.fl_str_mv application/pdf
dc.publisher.none.fl_str_mv Elsevier
publisher.none.fl_str_mv Elsevier
dc.source.none.fl_str_mv reponame:RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
instname:Universitat Politècnica de València (UPV)
instname_str Universitat Politècnica de València (UPV)
reponame_str RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
collection RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
repository.name.fl_str_mv
repository.mail.fl_str_mv
_version_ 1869414683211988992
score 15,300719

Similares en LA Referencia