Characterizing and Detecting Hateful Users on Twitter

By Manoel Horta Ribeiro, Pedro H. Calais, Yuri A. Santos, Virgílio A. F. Almeida, Wagner Meira Jr. 2018

287 citations recorded by the source corpus.

Current approaches to characterize and detect hate speech focus on content posted in Online Social Networks (OSNs). They face shortcomings to get the full picture of hate speech due to its subjectivity and the noisiness of OSN text. This work partially addresses these issues by shifting the focus towards users. We obtain a sample of Twitter's retweet graph with 100,386 users and annotate 4,972 as hateful or normal, and also find 668 users suspended after 4 months. Our analysis shows that hateful/suspended users differ from normal/active ones in terms of their activity patterns, word usage and network structure. Exploiting Twitter's network of connections, we find that a node embedding algorithm outperforms content-based approaches for detecting both hateful and suspended users. Overall, we present a user-centric view of hate speech, paving the way for better detection and understanding of this relevant and challenging issue.

Subfields: Artificial Intelligence, Communication, 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

In Communication

In Natural Language Processing

In Safety, Robustness & Interpretability

Citations in the corpus

The Risk of Racial Bias in Hate Speech Detection (2019, background): “In addition to DWMW 17 and FDCL 18, published Twitter corpora include Golbeck et al. (2017), which uses a somewhat restrictive deﬁnition of abuse, and Ribeiro et al. (2018), which is focused on network features, rather than text.”
The Hateful Memes Challenge: Detecting Hate Speech in Multimodal Memes (2020, background): “Hate speech There has been a lot of work in recent years on detecting hate speech in network science [61] and natural language processing [81, 65, 19].”
HateXplain: A Benchmark Dataset for Explainable Hate Speech Detection (2020, methodology): “To tackle this, various methods have been proposed for hate speech detection (Burnap and Williams 2016; Ribeiro et al. 2018; Zhang, Robinson, and Tepper 2018; Qian et al. 2018a).”
Resources and benchmark corpora for hate speech detection: a systematic review (2020, background): “To this effect, recent work is exploring the alternative route of collecting data for HS detection from ‘‘hateful’’ users (Ribeiro et al. 2018; Mishra et al. 2018).”
TweetBLM: A Hate Speech Dataset and Analysis of Black Lives Matter-related Microblogs on Twitter (2020, background): “Most commonly available hate speech classiﬁcation datasets are comprise binary classes (hate and non-hate) (Davidson et al., 2017), (Gao and Huang, 2017), (Ribeiro et al., 2018), (Qian et al., 2019).”
Tackling Online Abuse: A Survey of Automated Abuse Detection Methods (2019, background): “Ribeiro et al. (2018) and Mishra et al. (2019) applied graph neural networks ( Kipf and Welling, 2017; Hamilton et al., 2017) to social graphs in order to generate user embeddings (i.e., proﬁles) that capture not only their surrounding community but also their linguistic behavior.”
I Feel Offended, Don’t Be Abusive! Implicit/Explicit Messages in Offensive and Abusive Language (2020, background): “Some work in this direction has already been done (Ribeiro et al., 2018; Mishra et al., 2018; Wiegand et al., 2018b).”
Directions in Abusive Language Training Data: Garbage In, Garbage Out (2020, background): “49 of the training datasets are annotated at the level of the post, one dataset is annotated at the level of the user [88], and none of them are annotated at the level of the comment thread.”
Directions in abusive language training data, a systematic review: Garbage in, garbage out (2020, background): “Of the 63 training datasets, 62 are annotated at the level of the post and only one dataset is annotated at the level of the user [74].”
Towards multidomain and multilingual abusive language detection: a survey (2021, unknown): “com/manoelhortaribeiro/HatefulUsersTwitter [114]”
Predicting Anti-Asian Hateful Users on Twitter during COVID-19 (2021, background): “Among a few, Almerekhi et al. (2020) identify linguistic markers that trigger toxicity in online discussions, Ribeiro et al. (2018) detect the users who express online hate by checking their profiles or linguistic characteristics, Qian et al. (2018) incorporate users’ historical tweets and user…”
Spread of Hate Speech in Online Social Media (2018, methodology): “In (Ribeiro et al. 2018a), the authors study the user characteristics of hateful accounts on Twitter and found that the hateful user accounts differ significantly from normal user accounts on the basis of activity, network centrality, and the type of content they produce.”