ELF22: A Context-based Counter Trolling Dataset to Combat Internet Trolls

Proceedings of the 13th Conference on Language Resources and Evaluation (LREC 2022), pages 3530–3541 Marseille, 20-25 June 2022 © European Language Resources Association (ELRA), licensed under CC-BY-NC-4.0 3530 ELF22: A Context-based Counter Trolling Dataset to Combat Internet Trolls Huije Lee∗, Young Ju NA∗†, Hoyun Song, Jisu Shin, Jong C. Park KAIST Daehak-ro 291, Yuseong-gu, Daejeon, Republic of Korea {jae4258,hysong,jsshin,park}@nlp.kaist.ac.kr youngju073092@gmail.com Abstract Online trolls increase social costs and cause psychological damage to individuals. With the proliferation of automated accounts making use of bots for trolling, it is difficult for targeted individual users to handle the situation both quantitatively and qualitatively. To address this issue, we focus on automating the method to counter trolls, as counter responses to combat trolls encourage community users to maintain ongoing discussion without compromising freedom of expression. For this purpose, we propose a novel dataset for automatic counter response generation. In particular, we constructed a pair-wise dataset that includes troll comments and counter responses with labeled response strategies, which enables models fine-tuned on our dataset to generate responses by varying counter responses according to the specified strategy. We conducted three tasks to assess the effectiveness of our dataset and evaluated the results through both automatic and human evaluation. In human evaluation, we demonstrate that the model fine-tuned on our dataset shows a significantly improved performance in strategy-controlled sentence generation. Keywords: Natural language, countering trolls, conditional text generation 1. Introduction The International Telecommunication Union (ITU) stated that the number of internet users increased from 4.1 billion in 2019 to 4.9 billion in 2021, as 782 mil- lion people started to participate in cyberspace activities during the COVID-19 pandemic. The Pew Research Center in 2021 reported that 93% of American adults participate in internet communities to interact with oth- ers. Together with the positive side of online communi- ties that promote open discussion and facilitate social movements, the number of users experiencing online trolling or abuse has also increased. The Australia Insti- tute (2019) reported the estimated social cost of online trolling and abuse at $3.7 billion in 2019. Online trolling is a malicious attempt to provoke strong reactions from an interlocutor. Trolls, it would seem, do not care what they are saying as long as they have an impact on the chosen target (Fichman and Sanfilippo, 2016; Mihaylov and Nakov, 2016; Golf-Papez and Veer, 2017). Regarding the issue of online trolls, there have been attempts by community managers such as Twitter, Facebook, and Reddit to delete troll comments or by users to ignore them (Sibai et al., 2015). However, compared to the troll attackers, the voice of the victim group is overall weak. For example, Asian hate speech related to COVID-19 showed a large difference in the number of tweets, as there were 3 times more users displaying hate speech than those displaying counter speech, and hateful bots were more active and more *equal contribution †Present Address: Young Ju NA, Universit´e Sorbonne Nouvelle, 17 rue de Santeuil 75005 Paris successful in attracting followers than counter-hate bots (Ziems et …

Content truncated to 500 words