Online Communication of eSports Viewers: Topic Modeling Approach
This paper represents a brief overview of the communication of eSports viewers during the broadcasts of Dota 2 tournaments on the streaming platform Twitch.tv. We employed a topic modelling algorithm Twitter-LDA to analyse the contents of chat messages. We found that different stages of the stream can trigger different ways of behaviour depending on the type of the content. Game sessions more often had short emotional expressions, while breaks and moments of game inactivity included more analytical discussions and sociability.