Incivility Detection in Online Comments

Proceedings of the Eighth Joint Conference on Lexical and Computational Semantics (*SEM), pages 283–291 Minneapolis, June 6–7, 2019. c⃝2019 Association for Computational Linguistics 283 Incivility Detection in Online Comments Farig Sadeque School of Information University of Arizona Tucson, AZ 85721 farig@email.arizona.edu Stephen Rains Dept. of Communication University of Arizona Tucson, AZ 85721 srains@email.arizona.edu Yotam Shmargad School of Govt. and Public Policy University of Arizona Tucson, AZ 85721 yotam@email.arizona.edu Kate Kenski Dept. of Communication University of Arizona Tucson, AZ 85721 kkenski@email.arizona.edu Kevin Coe Dept. of Communication University of Utah Salt Lake City, UT 84112 kevin.coe@utah.edu Steven Bethard School of Information University of Arizona Tucson, AZ 85721 bethard@email.arizona.edu Abstract Incivility in public discourse has been a ma- jor concern in recent times as it can affect the quality and tenacity of the discourse negatively. In this paper, we present neural models that can learn to detect name-calling and vulgarity from a newspaper comment section. We show that in contrast to prior work on detecting toxic language, fine-grained incivilities like name- calling cannot be accurately detected by sim- ple models like logistic regression. We ap- ply the models trained on the newspaper com- ments data to detect uncivil comments in a Russian troll dataset, and find that despite the change of domain, the model makes accurate predictions. 1 Introduction Online harassment, colloquially known as cyber- bullying or cyber harassment, has been rampant since the introduction of the Internet to the general population. It has been a major cause of concern since the mid- and late-90’s, and is a thoroughly researched topic in the fields of social science, be- havioral science, network science and computer se- curity. Cyberbullying is a form of harassment that is carried out using electronic modes of communi- cation like computer, phone, and in almost all the cases in recent years, the Internet. Cyberbullying is defined as a “willful and repeated harm inflicted through the medium of electronic text” by Patchin and Hinduja (2006)- but this phenomenon goes far beyond the scope of just electronic text. A more comprehensive definition of cyberbullying can be found in one of their later works, where they de- fined cyberbullying as “a form of harassment using electronic mode of communication” (Hinduja and Patchin, 2008). Fauman (2008) described cyberbul- lying as “bullying through the use of technology such as the Internet and cellular phones”. The spectrum of online harassment is vast; hence, we focus on one segment of this phe- nomenon: online incivility. Incivility has been rampant in American society for quite some time. Incivility is described as features of discussion that convey an unnecessarily disrespectful tone toward the discussion forum, its participants, or its topics (Coe et al., 2014). While it is often said that in- civility is “very much in the eye of the beholder” and what is civil to someone may be uncivil to another (Kenski et al., 2017), some are universal nevertheless. One study has suggested that 69% of Americans believe that incivility in public dis- course has become a rampant …

Content truncated to 500 words