Transition between distribution patterns in human dynamics with high activity

Abstract

Complex interactions among a large number of individuals lead to multiple patterns of collective human behavior. However, the theoretical prediction of pattern transitions has not been empirically confirmed. This is because in previous empirical studies, different patterns were observed in different systems, and the coexistence of multiple patterns in the same system was rarely found. By investigating nearly 10 million messages in 252 QQ groups, we find rich distribution patterns of interevent time for human collective behavior, including the transitions between bimodal distribution, double-power-law distribution, and single-power-law distribution. The model developed in this paper suggests that human physiological rhythms and the high collective activity play key roles in the presentation of the single-power-law distribution. These results enhance the empirical research in the field of human dynamics, and are helpful for understanding many complex socioeconomic phenomena.