Speaker

Vinay Koshy

Abstract

Online communities are core to our digital lives. We mingle on social media like Reddit and Discord, share knowledge through platforms like StackExchange and Wikipedia, and use open-source software created by developer communities. Most online communities rely on content moderation policies to dictate the boundaries of acceptable speech. Yet community members have little say in how these moderation policies are created or enforced. In this talk, I will discuss my work in human-computer interaction on developing computational approaches for democratizing online community governance. I will first begin by arguing that the fluidity of online community membership creates unique dynamics that should inform the design of governance procedures. I will demonstrate how approaches from computer science can be used to study these dynamics. I will then discuss two computational approaches for eliciting and incorporating diverse perspectives into community moderation policy: an opinion polling framework that was operationalized in a real online community, and a machine learning-guided panel review system. I will conclude by outlining future avenues through which computational methods can be used to study online community dynamics and facilitate democratic governance.

Bio

Vinay Koshy is a computer science PhD student at the University of Illinois at Urbana-Champaign in the Social Spaces group, advised by Professor Karrie Karahalios. He applies natural language processing techniques to study online communities like Reddit subreddits and YouTube channels and builds tools to make the content moderation decisions in these communities more democratic. His work has won awards at top conferences in human-computer interaction, including a Best Paper award at CSCW and two Honorable Mention Awards at CHI.