mlsanityData to 14 Oct 2026

Hidden behind the obvious: misleading keywords and implicitly abusive language on social media

arXiv 2205.01374 ↗First posted 3 May 2022cs.CL cs.SIcited by 6

Not on the map. Why?

Abstract (arXiv)

While social media offers freedom of self-expression, abusive language carry significant negative social impact. Driven by the importance of the issue, research in the automated detection of abusive language has witnessed growth and improvement. However, these detection models display a reliance on strongly indicative keywords, such as slurs and profanity. This means that they can falsely (1a) miss abuse without such keywords or (1b) flag non-abuse with such keywords, and that (2) they perform poorly on unseen data. Despite the recognition of these problems, gaps and inconsistencies remain in the literature. In this study, we analyse the impact of keywords from dataset construction to model behaviour in detail, with a focus on how models make mistakes on (1a) and (1b), and how (1a) and (1b) interact with (2). Through the analysis, we provide suggestions for future research to address all three problems.

Key facts. Cited by 6 papers in the map (data through 14 Oct 2026). Builds on 28 of its 66 references that are in the map. First posted on arXiv 3 May 2022.

Builds on

28 of 66 references.

References · 28

  1. 2018BERT: Pre-training of Deep Bidirectional Transformers for Language Understandingcited by 40,399
  2. 2017Automated Hate Speech Detection and the Problem of Offensive Languagecited by 736
  3. 2019Social Bias Frames: Reasoning about Social and Power Implications of Languagecited by 380
  4. 2019Predicting the Type and Target of Offensive Posts in Social Mediacited by 271
  5. 2018Large Scale Crowdsourcing and Characterization of Twitter Abusive Behaviorcited by 270
  6. 2021Latent Hatred: A Benchmark for Understanding Implicit Hate Speechcited by 217
  7. 2019Racial Bias in Hate Speech and Abusive Language Detection Datasetscited by 209
  8. 2019SemEval-2019 Task 6: Identifying and Categorizing Offensive Language in Social Media (OffensEval)cited by 207
  9. 2018Hate Speech Dataset from a White Supremacy Forumcited by 198
  10. 2020SemEval-2020 Task 12: Multilingual Offensive Language Identification in Social Media (OffensEval 2020)cited by 138
  11. 2017Understanding Abuse: A Typology of Abusive Language Detection Subtaskscited by 116
  12. 2019A BERT-Based Transfer Learning Approach for Hate Speech Detection in Online Social Mediacited by 96
16 more
  1. 2021Towards generalisable hate speech detection: a review on obstacles and solutionscited by 66
  2. 2018All You Need is "Love": Evading Hate-speech Detectioncited by 56
  3. 2018Challenges for Toxic Comment Classification: An In-Depth Error Analysiscited by 56
  4. 2020SOLID: A Large-Scale Semi-Supervised Dataset for Offensive Language Identificationcited by 53
  5. 2017Detecting Hate Speech in Social Mediacited by 42
  6. 2020Stereotypical Bias Removal for Hate Speech Detection Task using Knowledge-based Generalizationscited by 40
  7. 2020Detecting East Asian Prejudice on Social Mediacited by 40
  8. 2021Overview of the HASOC track at FIRE 2020: Hate Speech and Offensive Content Identification in Indo-European Languagescited by 38
  9. 2018Hate Speech Detection: A Solved Problem? The Challenging Case of Long Tail on Twittercited by 36
  10. 2019Tackling Online Abuse: A Survey of Automated Abuse Detection Methodscited by 33
  11. 2018Leveraging Intra-User and Inter-User Representation Learning for Automated Hate Speech Detectioncited by 25
  12. 2020On Cross-Dataset Generalization in Automatic Detection of Online Abusecited by 20
  13. 2017Recognizing Explicit and Implicit Hate Speech Using a Weakly Supervised Two-path Bootstrapping Approachcited by 17
  14. 2018Degree based Classification of Harmful Speech using Twitter Datacited by 8
  15. 2019Modeling Islamist Extremist Communications on Social Media using Contextual Dimensions: Religion, Ideology, and Hatecited by 4
  16. 2020Duluth at SemEval-2019 Task 6: Lexical Approaches to Identify and Categorize Offensive Tweetscited by 4

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2022: 1; 2023: 1; 2024: 2; 2025: 1; 2026: 1.

6 papers.

  1. 2023SemEval-2023 Task 10: Explainable Detection of Online Sexismcited by 48
  2. 2024MentalManip: A Dataset For Fine-grained Analysis of Mental Manipulation in Conversationscited by 21
  3. 2024Harmful Speech Detection by Language Models Exhibits Gender-Queer Dialect Biascited by 16
  4. 2022Combating high variance in Data-Scarce Implicit Hate Speech Classificationcited by 2
  5. 2025Cracking the Code: Enhancing Implicit Hate Speech Detection through Coding Classificationcited by 2
  6. 2026IYKYK (But AI Doesn't): Automated Content Moderation Does Not Capture Communities' Heterogeneous Attitudes Towards Reclaimed Languagecited by 0