A Dataset of Offensive Language in Kosovo Social Media

Proceedings of the 13th Conference on Language Resources and Evaluation (LREC 2022), pages 1860–1869 Marseille, 20-25 June 2022 © European Language Resources Association (ELRA), licensed under CC-BY-NC-4.0 1860 A Dataset of Offensive Language in Kosovo Social Media Adem Ajvazi, Christian Hardmeier IT University of Copenhagen Rued Langgaards Vej 7, 2300 København S ademajvazi22@gmail.com, chrha@itu.dk Abstract Social media are an important part of people’s lives. Unfortunately, many public social media spaces are rife with bullying and offensive language, creating an unsafe environment for their users. In this paper, we present a new dataset for offensive language detection in Albanian. The dataset is composed of user-generated comments on Facebook and YouTube from the channels of selected Kosovo news platforms. It is annotated according to the three levels of the OLID annotation scheme. We also show results of a baseline system for offensive language classification based on a fine-tuned BERT model and compare it with the Danish DKhate dataset, which is similar in scope and size. In a transfer learning setting, we find that merging the Albanian and Danish training sets leads to improved performance for prediction on Danish, but not Albanian, on both offensive language recognition and distinguishing targeted and untargeted offence. Keywords: offensive language, social media, Kosovo, Albanian 1. Introduction The usage of the internet continues to grow in recent years. Statistical data show that around 60% of the world’s population has an online presence. Most of these internet users use social media platforms. The analysis shows that their number is around 4 billion1. Simply put, this means that more people now use so- cial media than do not. These easily accessible plat- forms have given a voice to many individuals to share their stories (Swamy et al., 2019). However, the down- side is that these platforms can be misused to spread hate and offend other individuals. So, their high pop- ularity combined with the high number of people that post about their experiences and opinions has led to not only an exponential increase in the user-generated con- tent but to a massive increase of the offensive language as well. Even though forms of offensive language like bullying or hate speech have existed before social me- dia platforms, these have given their users the power to reach and affect the lives of billions of people. It has been reported that offensive language posted here has not only created mental and psychological distress to the users, but it also forced many to deactivate their ac- counts and even commit suicide in extreme cases (Ku- mar et al., 2018). This has become a serious concern for government organizations and for social media plat- forms themselves. As we aim to create online environments that are safe for all users, fighting offensive language is becoming more and more important in a world where online so- cial media plays a significant role in shaping people’s minds (Park and Fung, 2017), and social media giants such as Facebook, Instagram and Twitter have come under increased pressure to …

Content truncated to 500 words