Who will Subscribe to My Streaming Channel?: The Case of Twitch

Abstract

This paper investigates whether subscribers (paying viewers) can be identified by their chat messages in live streaming services, Twitch. Collecting data of twenty streamers on Twitch, this study conducts two studies: (i) presenting sentimental differences of viewers chat messages between unsubscribers (free viewers) and subscribers (paying viewers), and (ii) employing a set of machine learning-based classifiers which attempt to accurately identify subscribers. The results show that the machine learning-based classifiers with sentimental features provide the potentiality to identify paying viewers, suggesting their utility as a research toolkit.