HARALD: Augmenting Hate Speech Data Sets with Real Data

Tal Ilan; Dan Vilenchik

HARALD: Augmenting Hate Speech Data Sets with Real Data

Department of Communication Systems Engineering

Research output: Contribution to conference › Paper › peer-review

Abstract

The successful completion of the hate speech detection task hinges upon the availability of rich and variable labeled data, which is hard to obtain. In this work, we present a new approach for data augmentation that uses as input real unlabelled data, which is carefully selected from online platforms where invited hate speech is abundant. We show that by harvesting and processing this data (in an automatic manner), one can augment existing manually-labeled datasets to improve the classification performance of hate speech classification models. We observed an improvement in F1-score ranging from 2.7% and up to 9.5%, depending on the task (in- or cross-domain) and the model used.

Original language	English
Pages	2241-2248
Number of pages	8
State	Published - 1 Jan 2022
Event	2022 Findings of the Association for Computational Linguistics: EMNLP 2022 - Abu Dhabi, United Arab Emirates Duration: 7 Dec 2022 → 11 Dec 2022

Conference

Conference	2022 Findings of the Association for Computational Linguistics: EMNLP 2022
Country/Territory	United Arab Emirates
City	Abu Dhabi
Period	7/12/22 → 11/12/22

ASJC Scopus subject areas

Computational Theory and Mathematics
Computer Science Applications
Information Systems

Cite this

@conference{0b594505d5fc4a0dbe10b3e8b5423ffd,

title = "HARALD: Augmenting Hate Speech Data Sets with Real Data",

abstract = "The successful completion of the hate speech detection task hinges upon the availability of rich and variable labeled data, which is hard to obtain. In this work, we present a new approach for data augmentation that uses as input real unlabelled data, which is carefully selected from online platforms where invited hate speech is abundant. We show that by harvesting and processing this data (in an automatic manner), one can augment existing manually-labeled datasets to improve the classification performance of hate speech classification models. We observed an improvement in F1-score ranging from 2.7% and up to 9.5%, depending on the task (in- or cross-domain) and the model used.",

author = "Tal Ilan and Dan Vilenchik",

note = "Publisher Copyright: {\textcopyright} 2022 Association for Computational Linguistics.; 2022 Findings of the Association for Computational Linguistics: EMNLP 2022 ; Conference date: 07-12-2022 Through 11-12-2022",

year = "2022",

month = jan,

day = "1",

language = "English",

pages = "2241--2248",

}

TY - CONF

T1 - HARALD

T2 - 2022 Findings of the Association for Computational Linguistics: EMNLP 2022

AU - Ilan, Tal

AU - Vilenchik, Dan

PY - 2022/1/1

Y1 - 2022/1/1

N2 - The successful completion of the hate speech detection task hinges upon the availability of rich and variable labeled data, which is hard to obtain. In this work, we present a new approach for data augmentation that uses as input real unlabelled data, which is carefully selected from online platforms where invited hate speech is abundant. We show that by harvesting and processing this data (in an automatic manner), one can augment existing manually-labeled datasets to improve the classification performance of hate speech classification models. We observed an improvement in F1-score ranging from 2.7% and up to 9.5%, depending on the task (in- or cross-domain) and the model used.

AB - The successful completion of the hate speech detection task hinges upon the availability of rich and variable labeled data, which is hard to obtain. In this work, we present a new approach for data augmentation that uses as input real unlabelled data, which is carefully selected from online platforms where invited hate speech is abundant. We show that by harvesting and processing this data (in an automatic manner), one can augment existing manually-labeled datasets to improve the classification performance of hate speech classification models. We observed an improvement in F1-score ranging from 2.7% and up to 9.5%, depending on the task (in- or cross-domain) and the model used.

UR - http://www.scopus.com/inward/record.url?scp=85149819849&partnerID=8YFLogxK

M3 - Paper

SP - 2241

EP - 2248

Y2 - 7 December 2022 through 11 December 2022

ER -

HARALD: Augmenting Hate Speech Data Sets with Real Data

Abstract

Conference

ASJC Scopus subject areas

Other files and links

Fingerprint

Cite this