mlsanityData to 14 Oct 2026

Predicting Different Types of Subtle Toxicity in Unhealthy Online Conversations

arXiv 2106.03952 ↗First posted 7 Jun 2021cs.CLcited by 5

Not on the map. Why?

Abstract (arXiv)

This paper investigates the use of machine learning models for the classification of unhealthy online conversations containing one or more forms of subtler abuse, such as hostility, sarcasm, and generalization. We leveraged a public dataset of 44K online comments containing healthy and unhealthy comments labeled with seven forms of subtle toxicity. We were able to distinguish between these comments with a top micro F1-score, macro F1-score, and ROC-AUC of 88.76%, 67.98%, and 0.71, respectively. Hostile comments were easier to detect than other types of unhealthy comments. We also conducted a sentiment analysis which revealed that most types of unhealthy comments were associated with a slight negative sentiment, with hostile comments being the most negative ones.

Key facts. Cited by 5 papers in the map (data through 14 Oct 2026). Builds on 22 of its 69 references that are in the map. First posted on arXiv 7 Jun 2021.

Builds on

22 of 69 references.

References · 22

  1. 2018BERT: Pre-training of Deep Bidirectional Transformers for Language Understandingcited by 40,399
  2. 2013Efficient Estimation of Word Representations in Vector Spacecited by 4,375
  3. 2017Automated Hate Speech Detection and the Problem of Offensive Languagecited by 736
  4. 2014Distributed Representations of Sentences and Documentscited by 678
  5. 2017Understanding Abuse: A Typology of Abusive Language Detection Subtaskscited by 116
  6. 2018Challenges for Toxic Comment Classification: An In-Depth Error Analysiscited by 56
  7. 2019A Just and Comprehensive Strategy for Using NLP to Address Online Abusecited by 49
  8. 2019Data Augmentation via Dependency Tree Morphing for Low-Resource Languagescited by 45
  9. 2017Creating and Characterizing a Diverse Corpus of Sarcasm in Dialoguecited by 43
  10. 2020A Comparison of LSTM and BERT for Small Corpuscited by 40
  11. 2017A Network Perspective on Stratification of Multi-Label Datacited by 37
  12. 2016A Deeper Look into Sarcastic Tweets Using Deep Convolutional Neural Networkscited by 33
10 more
  1. 2016Mutual information for symmetric rank-one matrix estimation: A proof of the replica formulacited by 26
  2. 2018CASCADE: Contextual Sarcasm Detection in Online Discussion Forumscited by 26
  3. 2017A scikit-based Python environment for performing multi-label classificationcited by 24
  4. 2018Convolutional Neural Networks for Toxic Comment Classificationcited by 24
  5. 2019TalkDown: A Corpus for Condescension Detection in Contextcited by 21
  6. 2020Six Attributes of Unhealthy Conversationcited by 19
  7. 2020Don't Patronize Me! An Annotated Dataset with Patronizing and Condescending Language towards Vulnerable Communitiescited by 19
  8. 2018Forecasting the presence and intensity of hostility on Instagram using linguistic and social featurescited by 11
  9. 2017GPU-acceleration for Large-scale Tree Boostingcited by 8
  10. 2020Sarcasm Detection using Context Separators in Online Discoursecited by 7

Built on by

Cited by year (2026 to 14 Oct)

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

5 papers.

  1. 2024Toxicity Classification in Ukrainiancited by 7
  2. 2022ML_LTU at SemEval-2022 Task 4: T5 Towards Identifying Patronizing and Condescending Languagecited by 3
  3. 2026Position: Evaluation Scores Are Perishable Knowledge Claimscited by 3
  4. 2023How We Define Harm Impacts Data Annotations: Explaining How Annotators Distinguish Hateful, Offensive, and Toxic Commentscited by 2
  5. 2025Quantifying Feature Importance for Online Content Moderationcited by 0