Proactive Blocking through the Automated Identification of Likely Harassers
PDF

Supplementary Files

PDF

Keywords

Hindutva
online harassment
content moderation
Kashmir
activism
dissidents
blocking

Categories

How to Cite

Ifat Gazia, Trevor Hubbard, Timothy Scalona, Yena Kang, & Ethan Zuckerman. (2024). Proactive Blocking through the Automated Identification of Likely Harassers . Journal of Online Trust and Safety, 2(3). https://doi.org/10.54501/jots.v2i3.175

Abstract

Since people began interacting in computer-mediated spaces, there has been a need to block or silence abusive users. In 2014, Gamergate—a purported campaign for “ethics in game journalism,” which often seemed a misogynist protest against women in computer gaming—brought the issue of online harassment to popular attention and inspired a wave of tools and techniques to mitigate online abuse. Yet, it remains a serious problem. Individuals, particularly activists and political dissidents, can face intense harassment on platforms like X (formerly Twitter), designed to silence their speech. This paper proposes a method to block likely abusers on X, using Kashmiri dissidents and Hindutva (Hindu nationalist) harassers as a case study. We first interviewed six Kashmiri dissidents who use social media for their activism to better understand their unique online experiences. Then, using a combination of text analysis and social network analysis, based on a sampling of accounts provided by the interviewees, we developed a novel filtering method. Our tests indicate that the model is 97% effective at identifying accounts that were previously blocked for harassment. This model could be useful for screening interactions on X, and preemptively filtering any it identifies as potential harassment. While it may no longer be appropriate for protecting Kashmiri users—many of whom have fled social media platforms—this model could be used in other minority communities and on other social media platforms.

https://doi.org/10.54501/jots.v2i3.175
PDF
Creative Commons License

This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.

Copyright (c) 2024 Journal of Online Trust and Safety