Online communities are natural laboratories for how attention and influence shape what people share. We present an AI-supported approach that asks a simple question: why do some pieces of electronic word-of-mouth (eWOM) take off while others fade? Using 270 million posts from 1,145 Reddit communities in the video game domain, we combine topic modeling and sentiment analysis with maps of who replies to whom, creating a picture of exposure and attention. We then run simulation experiments that let messages travel through these networks and observe how far and how fast they spread. Two patterns stand out. First, where a person sits in the network—who sees them and through which paths—predicts downstream spread more reliably than whether the message is positive or negative. Second, in highly one-sided attention networks, simple cues such as how many others a user reaches can match or beat more complicated influence formulas. A focused look at Nintendo-related discussions confirms that “bridge” positions between clusters are especially powerful in kickstarting cascades. By linking everyday actions (replying and passing content along) to the structure of online communities, the study offers an intuitive, scalable account of social transmission in digital settings. The framework also supports ethical, early identification of high-impact members and emerging themes to encourage well-being and constructive engagement.

Authors:
Ivan Belik, Norwegian School of Economics, Norway
Denise Utochkin, Norwegian School of Economics, Norway

About the Presenter(s)
Ivan Belik is a Professor in the Department of Strategy and Management at Norwegian School of Economics. Interests: AI-enabled analytics, network analysis, digital innovation. Current project: AI-enhanced network analysis of social media communities

See this presentation on the full schedule – Wednesday Schedule

Bookmark