Comparative Analysis of Information Spreading Focused on Topics and Emotions via Temporal Point Process

SNS is used as an infrastructure that allows users to freely and frankly express their opinions, and a wide variety of contents are posted. Since it is free, it is often seen that controversial remarks are posted and other people’s posts are irresponsibly spread by methods such as retweeting. A survey by the Ministry of Internal Affairs and Communications in Japan reported that the criteria for whether or not to retweet were based on whether the content was interesting and whether the content sympathized with, rather than on the reliability of the source. In this study, we analyze whether there is a difference in diffusion on SNS depending on the topic, positive/negative, and the type of emotion of the posted contents. We classify such tweets by topic classification, positive-negative classification, and sentiment classification, and analyze the differences between the classified classes in terms of the number of retweets, the average posting interval, and the intensity of self-excitation and mutual excitation in Hawkes process. As a result of the analysis, we found significant differences in various indices among classes.