Understanding Abuse: A Typology of Abusive Language Detection Subtasks

By Zeerak Talat, Thomas Davidson, Dana Warmsley, Ingmar Weber. 2017

524 citations recorded by the source corpus.

As the body of research on abusive language detection and analysis grows, there is a need for critical consideration of the relationships between different subtasks that have been grouped under this label. Based on work on hate speech, cyberbullying, and online abuse we propose a typology that captures central similarities and differences between subtasks and discuss the implications of this for data annotation and feature construction. We emphasize the practical actions that can be taken by researchers to best approach their abusive language detection subtask of interest.

Subfields: Artificial Intelligence, Natural Language Processing, AI Ethics, Fairness & Society

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Natural Language Processing

In AI Ethics, Fairness & Society

Citations in the corpus

HateCheck: Functional Tests for Hate Speech Detection Models (2020, background): “as “implicit abuse” categories in previous research (Waseem et al., 2017; Zhang and Luo, 2019).”
Evaluating the Social Impact of Generative AI Systems in Systems and Society (2023, background): “Abusive and offensive language are a large umbrella for unsafe content, which can also include abuse and hate speech[151, 236].”
Recipes for Safety in Open-domain Chatbots (2020, methodology): “Waseem et al. (2017) advocate for partitioning abusive content according to what entity it is directed to, an approach adopted by the OLID/OffensEval datasets (Zampieri et al., 2019, 2020).”
The Hateful Memes Challenge: Detecting Hate Speech in Multimodal Memes (2020, background): “One issue is that not all of these works have agreed on what defines hate speech, and different terminology has been used, ranging from offensive or abusive language, to online harassment or aggression, to cyberbullying, to harmful speech, to hate speech [81].”
Learning from the Worst: Dynamically Generated Datasets to Improve Online Hate Detection (2021, background): “It is similar to the ‘implicit’ and ‘covert’ categories used in other taxonomies (Waseem et al., 2017; Vidgen and Yasseri, 2019; Kumar et al., 2018).”
Dynabench: Rethinking Benchmarking in NLP (2021, background): “Detecting hate is notoriously difﬁcult given the important role played by context and speaker (Leader May-nard and Benesch, 2016) and the variety of ways in which hate can be expressed (Waseem et al., 2017).”
Racial Bias in Hate Speech and Abusive Language Detection Datasets (2019, background): “The aim of these models is to identify abusive language that directly targets certain individuals or groups, particularly people belonging to protected categories (Waseem et al., 2017).”
Challenges and frontiers in abusive content detection (2019, methodology): “Waseem et al. suggest that one of the main differences between subtasks is whether content is ‘directed towards a specific entity or is directed towards a generalized group’ (Waseem et al., 2017).”
SemEval-2019 Task 6: Identifying and Categorizing Offensive Language in Social Media (OffensEval) (2019, background): “Given the multitude of terms and deﬁnitions used in the literature, recent studies have investigated common aspects of the abusive language detection sub-tasks (Waseem et al., 2017; Wiegand et al., 2018).”
Finding Microaggressions in the Wild: A Case for Locating Elusive Phenomena in Social Media Posts (2019, background): “Instead, much of the recent work has focused on explicitly toxic language (e.g., Waseem et al., 2017), with surveys of the area also overlooking this important and challenging task of recognizing this subtle toxicity (van Aken et al., 2018; Salminen et al., 2018; Fortuna and Nunes, 2018).”
Predicting the Type and Target of Offensive Posts in Social Media (2019, neutral): “Recently, Waseem et. al.”
Detection of Abusive Language: the Problem of Biased Datasets (2019, background): “One major distinction that has been proposed in the literature is the division into explicitly and implicitly abusive language (Waseem et al., 2017).”
Latent Hatred: A Benchmark for Understanding Implicit Hate Speech (2021, background): “Implicit hate speech is a subclass of hate speech deﬁned by the use of coded or indirect language such as sarcasm, metaphor and circumlocution to disparage a protected group or individual, or to convey prejudicial and harmful views about them (Gao et al., 2017; Waseem et al., 2017).”
SemEval-2020 Task 12: Multilingual Offensive Language Identification in Social Media (OffensEval 2020) (2020, background): “Given the multitude of terms and deﬁnitions used in the literature, a few recent studies have investigated the common aspects of different abusive language detection subtasks (Waseem et al., 2017; Wiegand et al., 2018).”
Resources and benchmark corpora for hate speech detection: a systematic review (2020, methodology): “Moreover, the involvement of the scientiﬁc community resulted in a number of evaluation tasks organized in different languages, releasing benchmark corpora and encouraging participants to develop their own classiﬁcation systems.”
HateBERT: Retraining BERT for Abusive Language Detection in English (2020, background): “This interest has resulted in a fragmented picture of the various language phenomena at stake accompanied by a variety of deﬁnitions and (in)compatibility of the annotations (Waseem et al., 2017).”
COLD: A Benchmark for Chinese Offensive Language Detection (2022, background): “Further, based on the target, Offensive is subdivided into Attack Individuals and Attack Groups following Waseem et al. (2017).”
A BERT-Based Transfer Learning Approach for Hate Speech Detection in Online Social Media (2019, unknown): “ing datasets [2,21,24].”
Studying Generalisability across Abusive Language Detection Datasets (2019, methodology): “Waseem et al. (2017b) proposed that there exists an overlap between these subtasks and subsequently proposed a typology that emphasises identifying the target of abuse and whether the abuse is implicit or explicit.”