DALC: the Dutch Abusive Language Corpus

Tommaso Caselli; Arjan Schelhaas; Marieke Weultjes; Folkert Leistra; Hylke van der Veen; Gerben Timmerman; Malvina Nissim

doi:10.18653/v1/2021.woah-1.6

DALC: the Dutch Abusive Language Corpus

Tommaso Caselli, Arjan Schelhaas, Marieke Weultjes, Folkert Leistra, Hylke van der Veen, Gerben Timmerman, Malvina Nissim

Computational Linguistics (CL)

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Academic › peer-review

123 Downloads (Pure)

Abstract

As socially unacceptable language become pervasive in social media platforms, the need for automatic content moderation become more pressing. This contribution introduces the Dutch Abusive Language Corpus (DALC v1.0), a new dataset with tweets manually an- notated for abusive language. The resource ad- dress a gap in language resources for Dutch and adopts a multi-layer annotation scheme modeling the explicitness and the target of the abusive messages. Baselines experiments on all annotation layers have been conducted, achieving a macro F1 score of 0.748 for binary classification of the explicitness layer and .489 for target classification.

Original language	English
Title of host publication	Proceedings of the 5th Workshop on Online Abuse and Harm
Editors	Aida Mostafazadeh Davani, Douwe Kiela, Mathias Lambert, Bertie Vidgen, Vinodkumar Prabhakaran, Zeerak Waseem
Publisher	Association for Computational Linguistics (ACL)
Pages	54-66
Number of pages	13
DOIs	https://doi.org/10.18653/v1/2021.woah-1.6
Publication status	Published - 27-Jul-2021

Keywords

language models
hate speech
offensive language

Access to Document

10.18653/v1/2021.woah-1.6Licence: CC BY

DALC: the Dutch Abusive Language CorpusFinal publisher's version, 253 KBLicence: CC BY

Handle.net

Persistent link

Dutch Abusive Language Corpus v1.0
Caselli, T. (Creator), Weultjes, M. (Creator), Schelhaas, A. (Creator), Leistra, F. (Creator), van der Veen, H. (Creator) & Robben, M. (Creator), GitHub, 2021
https://github.com/tommasoc80/DALC
Dataset
DALC - Dutch Abusive Language Corpus
Caselli, T. (Creator), Weultjes, M. (Creator), Schelhaas, A. (Creator), Leistra, F. (Creator), Van Der Veen, H. (Creator), Robben, M. (Creator), Timmerman, G. (Creator), Zwart, V. (Creator), Van Der Noord, R. (Creator), Gnezdilov, Z. (Creator) & Theodoridis, D. (Creator), DataverseNL, 23-Mar-2023
DOI: 10.34894/hoinl3
Dataset

Cite this

Caselli, T., Schelhaas, A., Weultjes, M., Leistra, F., van der Veen, H., Timmerman, G., & Nissim, M. (2021). DALC: the Dutch Abusive Language Corpus. In A. Mostafazadeh Davani, D. Kiela, M. Lambert, B. Vidgen, V. Prabhakaran, & Z. Waseem (Eds.), Proceedings of the 5th Workshop on Online Abuse and Harm (pp. 54-66). Association for Computational Linguistics (ACL). https://doi.org/10.18653/v1/2021.woah-1.6

@inproceedings{bfbfefcd5bd5429283456e9a190e0bd9,

title = "DALC: the Dutch Abusive Language Corpus",

abstract = "As socially unacceptable language become pervasive in social media platforms, the need for automatic content moderation become more pressing. This contribution introduces the Dutch Abusive Language Corpus (DALC v1.0), a new dataset with tweets manually an- notated for abusive language. The resource ad- dress a gap in language resources for Dutch and adopts a multi-layer annotation scheme modeling the explicitness and the target of the abusive messages. Baselines experiments on all annotation layers have been conducted, achieving a macro F1 score of 0.748 for binary classification of the explicitness layer and .489 for target classification.",

keywords = "language models, hate speech, offensive language",

author = "Tommaso Caselli and Arjan Schelhaas and Marieke Weultjes and Folkert Leistra and {van der Veen}, Hylke and Gerben Timmerman and Malvina Nissim",

year = "2021",

month = jul,

day = "27",

doi = "10.18653/v1/2021.woah-1.6",

language = "English",

pages = "54--66",

editor = "{Mostafazadeh Davani}, Aida and Douwe Kiela and Mathias Lambert and Bertie Vidgen and Vinodkumar Prabhakaran and Zeerak Waseem",

booktitle = "Proceedings of the 5th Workshop on Online Abuse and Harm",

publisher = "Association for Computational Linguistics (ACL)",

}

Caselli, T, Schelhaas, A, Weultjes, M, Leistra, F, van der Veen, H, Timmerman, G & Nissim, M 2021, DALC: the Dutch Abusive Language Corpus. in A Mostafazadeh Davani, D Kiela, M Lambert, B Vidgen, V Prabhakaran & Z Waseem (eds), Proceedings of the 5th Workshop on Online Abuse and Harm. Association for Computational Linguistics (ACL), pp. 54-66. https://doi.org/10.18653/v1/2021.woah-1.6

DALC: the Dutch Abusive Language Corpus. / Caselli, Tommaso; Schelhaas, Arjan; Weultjes, Marieke et al.
Proceedings of the 5th Workshop on Online Abuse and Harm. ed. / Aida Mostafazadeh Davani; Douwe Kiela; Mathias Lambert; Bertie Vidgen; Vinodkumar Prabhakaran; Zeerak Waseem. Association for Computational Linguistics (ACL), 2021. p. 54-66.

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Academic › peer-review

TY - GEN

T1 - DALC

T2 - the Dutch Abusive Language Corpus

AU - Caselli, Tommaso

AU - Schelhaas, Arjan

AU - Weultjes, Marieke

AU - Leistra, Folkert

AU - van der Veen, Hylke

AU - Timmerman, Gerben

AU - Nissim, Malvina

PY - 2021/7/27

Y1 - 2021/7/27

N2 - As socially unacceptable language become pervasive in social media platforms, the need for automatic content moderation become more pressing. This contribution introduces the Dutch Abusive Language Corpus (DALC v1.0), a new dataset with tweets manually an- notated for abusive language. The resource ad- dress a gap in language resources for Dutch and adopts a multi-layer annotation scheme modeling the explicitness and the target of the abusive messages. Baselines experiments on all annotation layers have been conducted, achieving a macro F1 score of 0.748 for binary classification of the explicitness layer and .489 for target classification.

AB - As socially unacceptable language become pervasive in social media platforms, the need for automatic content moderation become more pressing. This contribution introduces the Dutch Abusive Language Corpus (DALC v1.0), a new dataset with tweets manually an- notated for abusive language. The resource ad- dress a gap in language resources for Dutch and adopts a multi-layer annotation scheme modeling the explicitness and the target of the abusive messages. Baselines experiments on all annotation layers have been conducted, achieving a macro F1 score of 0.748 for binary classification of the explicitness layer and .489 for target classification.

KW - language models

KW - hate speech

KW - offensive language

U2 - 10.18653/v1/2021.woah-1.6

DO - 10.18653/v1/2021.woah-1.6

M3 - Conference contribution

SP - 54

EP - 66

BT - Proceedings of the 5th Workshop on Online Abuse and Harm

A2 - Mostafazadeh Davani, Aida

A2 - Kiela, Douwe

A2 - Lambert, Mathias

A2 - Vidgen, Bertie

A2 - Prabhakaran, Vinodkumar

A2 - Waseem, Zeerak

PB - Association for Computational Linguistics (ACL)

ER -

DALC: the Dutch Abusive Language Corpus

Abstract

Keywords

Access to Document

Handle.net

Fingerprint

Datasets

Dutch Abusive Language Corpus v1.0

DALC - Dutch Abusive Language Corpus

Cite this