Cross-domain and Cross-lingual Abusive Language Detection: A Hybrid Approach with Deep Learning and a Multilingual Lexicon

Abstract

The development of computational methods to detect abusive language in social media within variable and multilingual contexts has recently gained significant traction. The growing interest is confirmed by the large number of benchmark corpora for different languages developed in the latest years. However, abusive language behaviour is multifaceted and available datasets are featured by different topical focuses. This makes abusive language detection a domain-dependent task, and building a robust system to detect general abusive content a first challenge. Moreover, most resources are available for English, which makes detecting abusive language in low-resource languages a further challenge. We address both challenges by considering ten publicly available datasets across different domains and languages. A hybrid approach with deep learning and a multilingual lexicon to cross-domain and cross-lingual detection of abusive content is proposed and compared with other simpler models. We show that training a system on general abusive language datasets will produce a cross-domain robust system, which can be used to detect other more specific types of abusive content. We also found that using the domain-independent lexicon HurtLex is useful to transfer knowledge between domains and languages. In the cross-lingual experiment, we demonstrate the effectiveness of our jointlearning model also in out-domain scenarios.

Resources and benchmark corpora for hate speech detection: a systematic review

“…For example, our survey captures a great availability of benchmark datasets for the evaluation of abusive language and hate speech detection systems, in several languages and with several topical focuses. This adds to the challenge of investigating architectures which are stable and well-performing across different languages and abusive domains, making it a more and more promising topic to research (Corazza et al 2020;Pamungkas and Patti 2019;Ousidhoum et al 2019).…”

Time of Your Hate: The Challenge of Time in Hate Speech Detection on Social Media

“…This evaluation gap is being bridged recently by evaluation campaigns for English, Spanish (SemEval [10]), German [11], and Italian (EVALITA [12]), whose shared tasks released annotated datasets for hate speech detection. The availability of benchmarks for system evaluation and datasets for hate speech detection in different languages made the challenge of investigating architectures, which are also stable and well-performing across different languages, an exciting issue to research [13,14].…”

To BAN or Not to BAN: Bayesian Attention Networks for Reliable Hate Speech Detection

“…Detecting abusive language for less-resourced languages is complex, and has inspired research in multilingual and cross-lingual methods [16]. These methods are especially useful when the involved languages are morphologically or geographically close [18]. In our work, we investigate hate speech detection methods for English, Croatian, and Slovene languages.…”