Dynamic commentary generation using terminology comprehension estimation based on viewer operations in live game videos

Abstract

The popularity of live game commentary, in which players themselves comment on their own play as they play a game, is growing. However, understanding the technical terminology is difficult for viewers who have never played the game. In this study, we created a system with the aim of providing viewing support and improving the viewing experience by generating dynamic explanations about technical terms based on the viewer's understanding level estimated from viewer operations. The results of evaluation experiments showed that dynamic explanations are desirable to viewers and contribute to viewers' understanding of technical terms better than existing viewers.