# A Survey on Automatic Online Hate Speech Detection in Low-Resource Languages \SetBgContents This work is shared under a \hrefhttps://creativecommons.org/licenses/by-sa/4.0/CC BY-SA 4.0 license unless otherwise noted Arpita Dutta2Kingshuk Roy3Abir Mondal4Arnab Mukhopadhyay5 (1Bennett University, India susmita.das@bennett.edu.in 2Techno Main Salt Lake, India adutta.7777@gmail.com 3Indian Institute of Information Technology & Management Gwalior, India rkingshuk2001@gmail.com 4Indian Institute of Engineering Science and Technology, Shibpur, India abir22197072@gmail.com 5National Institute of Technology, Silchar, India arnabmukhopadhyay76@gmail.com ) ###### Abstract The expanding influence of social media platforms over the past decade has impacted the way people communicate. The level of obscurity provided by social media and easy accessibility of the internet has facilitated the spread of hate speech. The terms and expressions related to hate speech gets updated with changing times which poses an obstacle to policy-makers and researchers in case of hate speech identification. With growing number of individuals using their native languages to communicate with each other, hate speech in these low-resource languages are also growing. Although, there is awareness about the English-related approaches, much attention have not been provided to these low-resource languages due to lack of datasets and online available data. This article provides a detailed survey of hate speech detection in low-resource languages around the world with details of available datasets, features utilized and techniques used. This survey further discusses the prevailing surveys, overlapping concepts related to hate speech, research challenges and opportunities. Keywords: article, template, simple ## 1 Introduction In the age of social networking and the exponential growth of social media platforms, people have gained freedom of speech and online anonymity. This has propelled the transmission of hate speech *[[65](https://arxiv.org/html/2411.19017v1#bib.bib65 "")]* on multiple social networking platforms and has presently transformed into a global issue. Social media platforms have become a centre for online communications, discussions, opinion exchange and arguments. People’s inclination to defend their opinions or ideologies often leads to online debates which occasionally transform into aggressive online behaviour and hate speech. Users often express their animosity in the form of hate speech by berating a certain individual or a group based on their religion, ethnicity, nationality, gender, sex, skin colour or even sexual orientation*[[135](https://arxiv.org/html/2411.19017v1#bib.bib135 "")]*. Social media companies have implemented their own policies for prevention of hate speech in the cyber-space which often relies on multiple users reporting a certain post or the platform admins regulating the timelines. The evolving social networks depicts increase in prejudiced perceptions, increase speed of hate speech spread*[[95](https://arxiv.org/html/2411.19017v1#bib.bib95 "")]*, delay in restrictions have posed a great challenge in counteracting against hate speech. Online social networks produce huge amounts of data which makes manual monitoring for hate speech infeasible. This naturally leads to requirement for automatic online hate speech detection techniques. As Natural Language Processing(NLP), Machine Learning(ML) and Deep Learning(DL) approaches make great headway in multiple pragmatic solutions to real-life problems, researchers obtained motivation in analysing different methods for detecting online hate speech. Researchers have assembled multiple large-scale datasets for hate speech, mostly in English language(the most prevalent language worldwide). Although several studies have been organized to deal with hate speech in multiple non-English languages, lack of resources has restricted extensive research. There are various other limitations apart from resource availability which have been acknowledged and require further research. Expanding on the previously available surveys on online hate speech, this paper provides an updated survey on automatic hate speech detection in low-resource languages. In this survey study, we have discussed the following issues associated with online hate speech detection, that collectively constitute our primary contributions: 1. 1. Various categories of hate speech and their corresponding background information along with multiple approaches for online hate speech detection are explained elaborately from the existing literature. 2. 2. The datasets for hate speech detection are studied well. Moreover, the available datasets for several low resource languages have received special focus. 3. 3. The multiple cutting-edge natural language processing and deep learning-based techniques used in low-resource languages are analyzed to identify online hate speech. 4. 4. The challenges encountered in identifying hate speech along with potential directions for further research have been addressed. ![Refer to caption](https://arxiv.org/html/extracted/6028218/fig/bb.png) *Figure 1: Survey Overview* The principal objective of this survey is to provide a comprehensive idea about the history of research on hate speech and a detailed account of its application in case of low-resource languages. This survey paper is organized as follows: Section 2222 consists background study detailing the definition of hate speech as explained by different social media giants, categorizing hate speech and the overlapping concepts related to hate speech. Section 3333 explains the search for related works and literature. Section 4444 delves into the datasets available in English and low-resource languages. Section 5555 expounds on automatic hate speech detection, primarily in low-resource languages. These low-resource languages have been presented continent-wise such as European, African, Latin American and Asian to bring clarity. Asian Languages have again been elaborated into Indic Languages and non-Indic languages as India has considerable spoken languages. Section 6666 details the research challenges and other opportunities with Section 7777 consisting of the Conclusion. The overview of this survey article has been depicted in Fig. [1](https://arxiv.org/html/2411.19017v1#S1.F1 "Figure 1 ‣ 1 Introduction ‣ A Survey on Automatic Online Hate Speech Detection in Low-Resource Languages"). ## 2 Background Study Although hate speech has existed in society for a prolonged time, detection of hate speech from the perspective of computer science is quite a recent development. ### 2.1 Hate Speech Defining hate speech can be quite difficult as it takes into account complex relationships and viewpoints among different groups of people and the definition is not consistent. As of now, the international human rights law does not have a comprehensive definition of hate speech, but UN Strategy and Plan of Action on Hate Speech has defined hate speech as follows: “any kind of commun

Chaitanya Pandey

Summary

Education

United Group of Institutions2024–2028