# Accurate and Data-Efficient Toxicity Prediction when Annotators Disagree

arXiv: https://arxiv.org/abs/2410.12217 · HTML: https://mlsanity.com/paper/2410.12217/

## Key facts

- Cited by 3 papers in the map (data through 14 Oct 2026).
- Builds on 14 of its 22 references that are in the map.
- First posted on arXiv 16 Oct 2024.

## Builds on (most cited first, 14 of 14)

- [Language Models are Few-Shot Learners](https://mlsanity.com/paper/2005.14165/) (2020)
- [RoBERTa: A Robustly Optimized BERT Pretraining Approach](https://mlsanity.com/paper/1907.11692/) (2019)
- [Neural Collaborative Filtering](https://mlsanity.com/paper/1708.05031/) (2017)
- [Dealing with Disagreements: Looking Beyond the Majority Vote in Subjective Annotations](https://mlsanity.com/paper/2110.05719/) (2021)
- [Quantifying the Persona Effect in LLM Simulations](https://mlsanity.com/paper/2402.10811/) (2024)
- [When the Majority is Wrong: Modeling Annotator Disagreement for Subjective Tasks](https://mlsanity.com/paper/2305.06626/) (2023)
- [Conditional Learning of Fair Representations](https://mlsanity.com/paper/1910.07162/) (2019)
- [Designing Toxic Content Classification for a Diversity of Perspectives](https://mlsanity.com/paper/2106.04511/) (2021)
- ["What We Can't Measure, We Can't Understand": Challenges to Demographic Data Procurement in the Pursuit of Fairness](https://mlsanity.com/paper/2011.02282/) (2020)
- [Imitation Learning by Estimating Expertise of Demonstrators](https://mlsanity.com/paper/2202.01288/) (2022)
- [The Ecological Fallacy in Annotation: Modelling Human Label Variation goes beyond Sociodemographics](https://mlsanity.com/paper/2306.11559/) (2023)
- [You Are What You Annotate: Towards Better Models through Annotator Representations](https://mlsanity.com/paper/2305.14663/) (2023)
- [Proxy Fairness](https://mlsanity.com/paper/1806.11212/) (2018)
- [Preference Elicitation with Soft Attributes in Interactive Recommendation](https://mlsanity.com/paper/2311.02085/) (2023)

## Built on by (most cited first, 3 of 3)

- [Beyond Demographics: Fine-tuning Large Language Models to Predict Individuals' Subjective Text Perceptions](https://mlsanity.com/paper/2502.20897/) (2025)
- [When Does Demographic Information Help? Data and Modeling Regimes for Perspective-Aware Hate Speech Detection](https://mlsanity.com/paper/2605.27313/) (2026)
- [A Survey of Toxicity Detection and Mitigation Strategies for Multilingual Language Models](https://mlsanity.com/paper/2606.25380/) (2026)

## Abstract (arXiv)

When annotators disagree, predicting the labels given by individual annotators can capture nuances overlooked by traditional label aggregation. We introduce three approaches to predicting individual annotator ratings on the toxicity of text by incorporating individual annotator-specific information: a neural collaborative filtering (NCF) approach, an in-context learning (ICL) approach, and an intermediate embedding-based architecture. We also study the utility of demographic information for rating prediction. NCF showed limited utility; however, integrating annotator history, demographics, and survey information permits both the embedding-based architecture and ICL to substantially improve prediction accuracy, with the embedding-based architecture outperforming the other methods. We also find that, if demographics are predicted from survey information, using these imputed demographics as features performs comparably to using true demographic data. This suggests that demographics may not provide substantial information for modeling ratings beyond what is captured in survey responses. Our findings raise considerations about the relative utility of different types of annotator information and provide new approaches for modeling annotators in subjective NLP tasks.