Hate Speech Detection: A Solved Problem? The Challenging Case of Long Tail on Twitter

By Ziqi Zhang, Le Luo. 2018

329 citations recorded by the source corpus.

In recent years, the increasing propagation of hate speech on social media and the urgent need for effective counter-measures have drawn significant investment from governments, companies, and empirical research. Despite a large number of emerging, scientific studies to address the problem, the performance of existing automated methods at identifying specific types of hate speech - as opposed to identifying non-hate -is still very unsatisfactory, and the reasons behind are poorly understood. This work undertakes the first in-depth analysis towards this problem and shows that, the very challenging nature of identifying hate speech on the social media is largely due to the extremely unbalanced presence of real hateful content in the typical datasets, and the lack of unique, discriminative features in such content, both causing them to reside in the 'long tail' of a dataset that is difficult to discover. To address this issue, we propose novel Deep Neural Network structures serving as effective feature extractors, and explore the usage of background information in the form of different word embeddings pre-trained from unlabelled corpora. We empirically evaluate our methods on the largest collection of hate speech datasets based on Twitter, and show that our methods can significantly outperform state of the art, as they are able to obtain a maximum improvement of between 4 and 16 percentage points (macro-average F1) depending on datasets.

Subfields: Artificial Intelligence, Communication, Information Systems, Natural Language Processing, Safety, Robustness & Interpretability

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Guangxuan Xu
2. Paul Röttger
3. Helen Z. Margetts
4. Dong Nguyen
5. Zeerak Talat
6. Ralf Krestel
7. Julian Risch
8. Soroush Vosoughi

In Communication

1. Julian Risch
2. Ralf Krestel
3. Alexander Löser
4. Betty van Aken
5. Helen Z. Margetts
6. Dong Nguyen
7. Guangxuan Xu
8. Soroush Vosoughi

In Information Systems

1. Zeerak Talat
2. Dong Nguyen
3. Julian Risch
4. Helen Z. Margetts
5. Ralf Krestel
6. Alexander Löser
7. Betty van Aken

In Natural Language Processing

1. Guangxuan Xu
2. Helen Z. Margetts
3. Ralf Krestel
4. Julian Risch
5. Dong Nguyen
6. Soroush Vosoughi
7. Zeerak Talat
8. Alexander Löser

In Safety, Robustness & Interpretability

1. Zeerak Talat
2. Ralf Krestel
3. Julian Risch
4. Alexander Löser
5. Betty van Aken
6. Guangxuan Xu
7. Soroush Vosoughi
8. Ruibo Liu

Citations in the corpus

• HateCheck: Functional Tests for Hate Speech Detection Models (2020, methodology): “Within each functionality, we aim to use diverse vocabulary and syntax to minimise similarity between test cases, which Zhou et al. (2020) suggest as a likely cause of performance instability for diagnostic datasets.”
• Challenges for Toxic Comment Classification: An In-Depth Error Analysis (2018, methodology): “However, many classifiers suffer from insufficient variance in methods and training data and therefore often tend to fail on the long tail of real world data (Zhang and Luo, 2018).”
• Enhanced Offensive Language Detection Through Data Augmentation (2020, background): “Fortunately, recent years have witnessed substantial progress in computational methods to tackle the abusive detection problem, particularly on Twitter (Davidson et al. 2017; Zhang and Luo 2019).”