Keyword extraction and classification from TV program viewers' tweets

Abstract

In recent years, young people no longer watch television as much as they used to. The main reasons for this are some TV programs are very long or viewing time is limited for viewers. In recent years, so-called live-action tweets, which comment on contents while watching TV programs in real time using Twitter, have been actively performed. In this study, we propose a method for extracting key phrases related to event scenes of TV programs using TV program live tweets, and propose a scene search system, aiming at efficient TV program viewing. As a result of the experiment, the program contents were estimated with an error of about 5% to 10% with respect to the program time. In addition, the extracted key phrases were visualized for each event scene category by t-SNE algorithm.