XHate-999: Analyzing and Detecting Abusive Language Across Domains and Languages

• DOI: 10.18653/v1/2020.coling-main.559
• Publication Date: 2021-01-08T13:58:31Z
• Authors: Goran Glavaš; Mladen Karan; Ivan Vulić
• Keywords: 0202 electrical engineering, electronic engineering, information engineering; 02 engineering and technology; 01 natural sciences; 004; 0105 earth and related environmental sciences

We present XHate-999, a multi-domain and multilingual evaluation data set for abusive language detection. By aligning test instances across six typologically diverse languages, XHate-999 the first time allows disentanglement of domain transfer effects in conduct series domain- language-transfer experiments with state-of-the-art monolingual transformer models, setting strong baseline results profiling as comprehensive resource Finally, we show that language-adaption, via intermediate masked modeling on corpora target language, can lead to substantially improved detection zero-shot setups.