OffendES: A New Corpus in Spanish for Offensive Language Research

Proceedings of Recent Advances in Natural Language Processing, pages 1096–1108 Sep 1–3, 2021. https://doi.org/10.26615/978-954-452-072-4_123 1096 OffendES: A New Corpus in Spanish for Offensive Language Research Flor Miriam Plaza-del-Arco, Arturo Montejo-R´aez, L. Alfonso Ure˜na-L´opez and Mar´ıa-Teresa Mart´ın-Valdivia Department of Computer Science, Advanced Studies Center in ICT (CEATIC) Universidad de Ja´en, Campus Las Lagunillas, 23071, Ja´en, Spain {fmplaza, amontejo, laurena, maite}@ujaen.es Abstract Offensive language detection and analysis has become a major area of research in Natural Language Processing. The freedom of par- ticipation in social media has exposed online users to posts designed to denigrate, insult or hurt them according to gender, race, religion, ideology, or other personal characteristics. Fo- cusing on young influencers from the well- known social platforms of Twitter, Instagram, and YouTube, we have collected a corpus com- posed of 47,128 Spanish comments manually labeled on offensive pre-defined categories. A subset of the corpus attaches a degree of con- fidence to each label, so both multi-class clas- sification and multi-output regression studies are possible. In this paper, we introduce the corpus, discuss its building process, novelties, and some preliminary experiments with it to serve as a baseline for the research community. 1 Introduction Offensive language is defined as the text which uses hurtful, derogatory, or obscene terms made by one person to another person (Wiegand et al., 2019). Related terms in the literature are hate speech (Waseem and Hovy, 2016), cyberbully- ing (Rosa et al., 2019), toxic language (van Aken et al., 2018), aggression language (Kumar et al., 2018), or abusive language (Nobata et al., 2016). Although there are subtle differences in meaning, they are all compatible with the above general defi- nition. Due to the well-acknowledged rise in digital so- cial interactions, in particular on social media plat- forms, the amount of offensive language is also steadily growing. Unfortunately, this type of prej- udiced communication can be extremely harmful and could lead to negative psychological effects among online users, especially among young peo- ple, causing anxiety, harassment, and even suicide in extreme cases (Hinduja and Patchin, 2010). At the same time, this issue also implicates gov- ernments, online communities, and social media platforms. In order to help fight this problem, these stakeholders are continuously taking appropriate actions to implement laws and policies combating hate speech. For instance, since 2013 the Coun- cil of Europe has sponsored the ”No Hate Speech” movement1 seeking to mobilize young people to combat hate speech and promote human rights on- line. In May 2016, the European Commission reached an agreement with Facebook, Microsoft, Twitter, and YouTube to create a “Code of con- duct on countering illegal hate speech online”2. From 2018 to 2020, platforms such as Instagram, Snapchat, and TikTok adopted the Code. Accord- ing to a Spanish report in 2019 on the evolution of hate crimes in Spain3, threats, insults, and discrim- ination are counted as the most repeated criminal acts, with the Internet (54.9%) and social media (17.2%) as the most widely used media to commit these actions. To help achieve …

Content truncated to 500 words