Toxic, Hateful, Offensive or Abusive? What Are We Really Classifying? An Empirical Analysis of Hate Speech Datasets
By Paula Fortuna, Juan Soler, Leo Wanner. 2020
184 citations recorded by the source corpus.
Subfields: Natural Language Processing, Evaluation & Benchmarks
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Natural Language Processing
- Y-Lan Boureau
- Gavin Abercrombie
- Verena Rieser
- Dirk Hovy
- Emily Dinan
- Shannon L. Spruit
- A. S. Bergman
- Svetlana Kiritchenko
In Evaluation & Benchmarks
- Verena Rieser
- Isar Nejadgholi
- Amanda Cercas Curry
- Emily Dinan
- Y-Lan Boureau
- Dirk Hovy
- Shannon L. Spruit
- Gavin Abercrombie
Citations in the corpus
- Anticipating Safety Issues in E2E Conversational AI: Framework and Tooling (2021, background): “Several surveys cover approaches to this problem (Schmidt & Wiegand, 2017; Fortuna & Nunes, 2018; Vidgen et al., 2019), and there exist reviews of offensive language datasets (Fortuna et al., 2020; Vidgen & Derczynski, 2020).”
- SafetyKit: First Aid for Measuring Safety in Open-domain Conversational Systems (2022, background): “Offensive content is therefore an umbrella term encompassing toxicity, hate speech, and abusive language (Fortuna et al., 2020).”
- Confronting Abusive Language Online: A Survey from the Ethical and Human Rights Perspective (2020, background): “Fortuna et al. (2020) demonstrated that many different definitions are being used for equivalent concepts, which makes most publicly available datasets incompatible.”
- Handling Bias in Toxic Speech Detection: A Survey (2022, background): “The literature surveyed here is distributed among various forms of toxic speech, known to adopt diferent and sometimes misleading terms to refer to equivalent classes [33].”
- Introducing CAD: the Contextual Abuse Dataset (2021, background): “, 2019b)) means that most training datasets have used some form of purposive (or ‘directed’) sampling to ensure enough entries are in the positive class (Fortuna et al., 2020).”
- ConvAbuse: Data, Analysis, and Benchmarks for Nuanced Abuse Detection in Conversational AI (2021, background): “Most work on detecting harmful content such as offensiveness, toxicity, abuse, and hate speech (see Fortuna et al. (2020) for definitions), has focused on social media platforms, foremost Twitter (e.g. Ball-Burack et al., 2021; Basile et al., 2019; Cao and Lee, 2020; Davidson et al., 2017; For-tuna…”