Multilingual and Multi-Aspect Hate Speech Analysis

By Nedjma Djouhra Ousidhoum, Zizheng Lin, Hongming Zhang, Yangqiu Song, Dit-Yan Yeung. 2019

347 citations recorded by the source corpus.

Current research on hate speech analysis is typically oriented towards monolingual and single classification tasks. In this paper, we present a new multilingual multi-aspect hate speech analysis dataset and use it to test the current state-of-the-art multilingual multitask learning approaches. We evaluate our dataset in various classification settings, then we discuss how to leverage our annotations in order to improve hate speech detection and classification in general.

Subfields: Artificial Intelligence, Natural Language Processing, Safety, Robustness & Interpretability, Signal Processing

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

In Natural Language Processing

In Safety, Robustness & Interpretability

In Signal Processing

Citations in the corpus

Language Generation Models Can Cause Harm: So What Can We Do About It? An Actionable Survey (2022, background): “Compared to English, such resources for other languages are severely lacking (Ousidhoum et al., 2019a).”
Social Bias Frames: Reasoning about Social and Power Implications of Language (2019, background): “Most related to our work, Ousidhoum et al. (2019) create a multilingual dataset of 13k tweets annotated for fve different emotion- and toxicity-related aspects, including a 16-class variable representing social groups targeted.”
HateXplain: A Benchmark Dataset for Explainable Hate Speech Detection (2020, methodology): “We combined the lexicon set provided by Davidson et al. (2017), Ousidhoum et al. (2019), and Mathew et al. (2019a) to generate a single lexicon.”
Resources and benchmark corpora for hate speech detection: a systematic review (2020, background): “This adds to the challenge of investigating architectures which are stable and well-performing across different languages and abusive domains, making it a more and more promising topic to research (Corazza et al. 2020; Pamungkas and Patti 2019; Ousidhoum et al. 2019).”
TweetBLM: A Hate Speech Dataset and Analysis of Black Lives Matter-related Microblogs on Twitter (2020, background): “Multilingual (Ousidhoum et al., 2019), (Chung et al., 2019) and multimodal (Gomez et al., 2020) hate speech datasets also played a very crucial role in understanding the diversity and extent of hate speech across different languages and modality.”
Probing Toxic Content in Large Pre-Trained Language Models (2021, methodology): “We trained an LR classiﬁer on four relatively different English datasets (Davidson et al., 2017; Founta et al., 2018; Ousidhoum et al., 2019; Zampieri et al., 2019), four others in Arabic (Ousid-houm et al., 2020; Albadi et al., 2018; Mulki et al., 2019; Zampieri et al., 2020), and the only one we…”
Hybrid Emoji-Based Masked Language Models for Zero-Shot Abusive Language Detection (2020, background): “Then, to investigate whether our results can be generalised also when dealing with typologi-cally different languages, we will test our approach on additional abusive language datasets covering other languages (Ousidhoum et al., 2019; Zampieri et al., 2020).”
Towards multidomain and multilingual abusive language detection: a survey (2021, methodology): “For example, the proposed dataset by Ousidhoum et al. [89] introduces some classes, including hate speech , abusive , offensive , disrespectful , fearful , and normal .”
SOLID: A Large-Scale Semi-Supervised Dataset for Offensive Language Identification (2020, background): “Hate speech identiﬁcation is by far the most studied abusive language detection task (Ousidhoum et al., 2019; Chung et al., 2019; Mathew et al., 2021).”
Directions in abusive language training data, a systematic review: Garbage in, garbage out (2020, methodology): “In 27 no information is given about the identity of annotators [27]; in 24 datasets very limited information is given [73], such as whether the annotator is a native speaker of the language; and in just 12 cases is detailed information given [47].”
Misogyny Detection in Twitter: a Multilingual and Cross-Domain Study (2020, positive): “Finally, Ousidhoum, Lin, Zhang, Song, and Yeung (2019) conducted a multilingual experiment on hate speech detection in three languages (i.e., English, France, and Arabic) by using Sluice Network Ruder12, Bingel, Augenstein, and Søgaard (2017) and Babylon multilingual word embeddings Smith, Turban,…”
ConvAbuse: Data, Analysis, and Benchmarks for Nuanced Abuse Detection in Conversational AI (2021, background): “Figures 3 and 4 show the distribution of labels for abuse directness and the target of the abuse for our dataset, OLID (Zampieri et al., 2019a) and Ousidhoum et al. (2019).”
XHate-999: Analyzing and Detecting Abusive Language Across Domains and Languages (2020, neutral): “The HatEval shared task (Basile et al., 2019) spans only English and German, and other works (Steinberger et al., 2017; Sohn and Lee, 2019; Ousidhoum et al., 2019; Steimel et al., 2019; Stappen et al., 2020; Corazza et al., 2020, inter alia ) similarly target only major European languages such as…”
Just Say No: Analyzing the Stance of Neural Dialogue Generation in Offensive Contexts (2021, background): “…social media posts or comments while ignoring the context (Davidson et al., 2017; Xu et al., 2012; Zampieri et al., 2019; Rosenthal et al., 2020; Kumar et al., 2018; Garibo i Orts, 2019; Ousidhoum et al., 2019; Breitfeller et al., 2019; Sap et al., 2020; Hada et al., 2021; Barikeri et al., 2021).”
Measuring Harmful Sentence Completion in Language Models for LGBTQIA+ Individuals (2022, background): “Few recent works covered hate speech on the basis of sexual orientation (Ousidhoum et al., 2019; Mollas et al., 2022; Kennedy et al., 2022; Chakravarthi et al., 2022; Nozza, 2022).”