HENIN: Learning Heterogeneous Neural Interaction Networks for Explainable Cyberbullying Detection on Social Media

Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing, pages 2543–2552, November 16–20, 2020. c⃝2020 Association for Computational Linguistics 2543 HENIN: Learning Heterogeneous Neural Interaction Networks for Explainable Cyberbullying Detection on Social Media Hsin-Yu Chen Institute of Data Science National Cheng Kung University Tainan, Taiwan d0107330@gmail.com Cheng-Te Li Institute of Data Science National Cheng Kung University Tainan, Taiwan chengte@ncku.edu.tw Abstract In the computational detection of cyberbully- ing, existing work largely focused on build- ing generic classifiers that rely exclusively on text analysis of social media sessions. De- spite their empirical success, we argue that a critical missing piece is the model explainabil- ity, i.e., why a particular piece of media ses- sion is detected as cyberbullying. In this pa- per, therefore, we propose a novel deep model, HEterogeneous Neural Interaction Networks (HENIN), for explainable cyberbullying detec- tion. HENIN contains the following compo- nents: a comment encoder, a post-comment co-attention sub-network, and session-session and post-post interaction extractors. Exten- sive experiments conducted on real datasets exhibit not only the promising performance of HENIN, but also highlight evidential com- ments so that one can understand why a media session is identified as cyberbullying. 1 Introduction In recent years, cyberbullying has become one of the most pressing online risks among youth and raised serious concerns in society. Cyberbullying is commonly defined as the electronic transmission of insulting or embarrassing comments, photos or videos, as illustrated in Figure 1. Harmful bully- ing behavior can include posting rumors, threats, pejorative labels, and sexual remarks. Research from the American Psychological Association and the White House has revealed more than 40% of young people in the US indicate that they have been bullied on social media platforms (Dinakar et al., 2012). Such a growing prevalence of cyberbully- ing on social media has detrimental societal effects, such as victims may experience lower self-esteem, increased suicidal ideation, and a variety of neg- ative emotional responses (Hinduja and Patchin, 2014). Therefore, it has become critically impor- tant to be able to detect and prevent cyberbullying UserID Comments Time 1 ... C1 ... 2019‐12‐09 09:23 2 ... C2 ... 2019‐12‐09 12:44 1 ... C3 ... 2019‐12‐09 18:05 3 ... C4 ... 2019‐12‐10 11:27 2 ... C5 ... 2019‐12‐10 20:51 5 ... C6 ... 2019‐12‐10 23:19 text, image, video A Post UserID: 0 Sequence of its comments insult! cyberbullying clues Figure 1: An illustration of a media session contain- ing an image/video/posted text and a sequence of com- ments. A cyberbullying session is typically composed of multiple insulting comments. on social media. Research in computer science aimed at identifying, predicting, and ultimately pre- venting cyberbullying through better understanding the nature and key characteristics of online cyber- bullying. In the literature, existing efforts toward auto- matically detecting cyberbullying have primarily focused on textual analysis of user comments, in- cluding keywords (Dadvar et al., 2012; Nahar et al., 2013; Nand et al., 2016) and sentiments analy- sis (Dani et al., 2017). These studies attempt to build a generic binary classifier by taking …

Content truncated to 500 words