# A little goes a long way: Improving toxic language classification despite data scarcity

arXiv: https://arxiv.org/abs/2009.12344 · HTML: https://mlsanity.com/paper/2009.12344/

## Key facts

- Cited by 12 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 25 Sep 2020.

## Built on by (most cited first, 12 of 12)

- [Towards Zero-Label Language Learning](https://mlsanity.com/paper/2109.09193/) (2021)
- [HateMM: A Multi-Modal Dataset for Hate Video Classification](https://mlsanity.com/paper/2305.03915/) (2023)
- [Detecting Harmful Content On Online Platforms: What Platforms Need Vs. Where Research Efforts Go](https://mlsanity.com/paper/2103.00153/) (2021)
- [A Survey of Methods for Addressing Class Imbalance in Deep-Learning Based Natural Language Processing](https://mlsanity.com/paper/2210.04675/) (2022)
- [Fairness-aware Class Imbalanced Learning](https://mlsanity.com/paper/2109.10444/) (2021)
- [ToxiCraft: A Novel Framework for Synthetic Generation of Harmful Information](https://mlsanity.com/paper/2409.14740/) (2024)
- [Dynamic Normativity: Necessary and Sufficient Conditions for Value Alignment](https://mlsanity.com/paper/2406.11039/) (2024)
- [Text Grafting: Near-Distribution Weak Supervision for Minority Classes in Text Classification](https://mlsanity.com/paper/2406.11115/) (2024)
- [UPB at SemEval-2021 Task 5: Virtual Adversarial Training for Toxic Spans Detection](https://mlsanity.com/paper/2104.08635/) (2021)
- [Automated Annotation of Scientific Texts for ML-based Keyphrase Extraction and Validation](https://mlsanity.com/paper/2311.05042/) (2023)
- [A Target-Aware Analysis of Data Augmentation for Hate Speech Detection](https://mlsanity.com/paper/2410.08053/) (2024)
- [From Measurement Instruments to Data: Leveraging Theory-Driven Synthetic Training Data for Classifying Social Constructs](https://mlsanity.com/paper/2410.12622/) (2024)

## Abstract (arXiv)

Detection of some types of toxic language is hampered by extreme scarcity of labeled training data. Data augmentation - generating new synthetic data from a labeled seed dataset - can help. The efficacy of data augmentation on toxic language classification has not been fully explored. We present the first systematic study on how data augmentation techniques impact performance across toxic language classifiers, ranging from shallow logistic regression architectures to BERT - a state-of-the-art pre-trained Transformer network. We compare the performance of eight techniques on very scarce seed datasets. We show that while BERT performed the best, shallow classifiers performed comparably when trained on data augmented with a combination of three techniques, including GPT-2-generated sentences. We discuss the interplay of performance and computational overhead, which can inform the choice of techniques under different constraints.