A Benchmark Dataset for Learning to Intervene in Online Hate Speech

Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing, pages 4755–4764, Hong Kong, China, November 3–7, 2019. c⃝2019 Association for Computational Linguistics 4755 A Benchmark Dataset for Learning to Intervene in Online Hate Speech Jing Qian†, Anna Bethke∗, Yinyin Liu∗, Elizabeth Belding†, William Yang Wang† † University of California, Santa Barbara ∗ Intel AI {jing qian,ebelding,william}@cs.ucsb.edu {anna.bethke,yinyin.liu}@intel.com Abstract Countering online hate speech is a critical yet challenging task, but one which can be aided by the use of Natural Language Processing (NLP) techniques. Previous research has primarily fo- cused on the development of NLP methods to automatically and effectively detect online hate speech while disregarding further action needed to calm and discourage individuals from using hate speech in the future. In addition, most existing hate speech datasets treat each post as an isolated instance, ignoring the conversational context. In this paper, we propose a novel task of generative hate speech intervention, where the goal is to au- tomatically generate responses to intervene during online conversations that contain hate speech. As a part of this work, we introduce two fully-labeled large-scale hate speech intervention datasets1 col- lected from Gab2 and Reddit3. These datasets pro- vide conversation segments, hate speech labels, as well as intervention responses written by Mechan- ical Turk4 Workers. In this paper, we also ana- lyze the datasets to understand the common inter- vention strategies and explore the performance of common automatic response generation methods on these new datasets to provide a benchmark for future research. 1 Introduction The growing popularity of online interactions through social media has been shown to have both positive and negative impacts. While social media improves information sharing, it also facilitates 1https://github.com/jing-qian/A-Benchmark-Dataset-for- Learning-to-Intervene-in-Online-Hate-Speech 2https://gab.ai 3https://www.reddit.com 4https://www.mturk.com Figure 1: An illustration of hate speech conversation between User 1 and User 2 and the interventions col- lected for our datasets. The check and the cross icons on the right indicate a normal post and a hateful post. The utterance following the human icon is a human- written intervention, while the utterance following the computer icon is machine-generated. the propagation of online harassment, including hate speech. These negative experiences can have a measurable negative impact on users. Recently, the Pew Research Center (Center, 2017) reported that “roughly four-in-ten Americans have person- ally experienced online harassment, and 63% con- sider it a major problem.” To address the growing problem of online hate, an extensive body of work has focused on devel- oping automatic hate speech detection models and datasets (Warner and Hirschberg, 2012; Waseem and Hovy, 2016; Davidson et al., 2017; Schmidt and Wiegand, 2017; ElSherief et al., 2018a,b; Qian et al., 2018a,b). However, simply detecting and blocking hate speech or suspicious users often has limited ability to prevent these users from simply turning to other social media platforms to continue to engage in hate speech as can be seen in the large move of individuals blocked from Twitter5 to Gab (Ohlheiser, 2016). What’s more, …

Content truncated to 500 words