Global Regularity and Individual Variability in Dynamic Behaviors of Human Communication
Jonathan J. H. Zhu,Tai-Quan Peng
DOI: https://doi.org/10.48550/arXiv.1911.01264
2019-11-04
Abstract:A new model, called "Human Dynamics", has been recently proposed that individuals execute activities based on a perceived priority of tasks, which can be characterized by a power-law distribution of waiting time between consecutive tasks (Barabasi, 2005). This power-law distribution has been found to exist in diverse human behaviors, such as mail correspondence, e-mail communication, webpage browsing, video-on-demand, and mobile phone calls. However, the pattern has been observed at the global (i.e., aggregated) level without considering individual differences. To guard against ecological fallacy, it is necessary to test the model at the individual level. The current study aims to address the following questions: Is the power-law uniform across individuals? What distribution do individual behaviors follow? We examine these questions with a client log file of nearly 4,000 Internet users' web browsing behavior and a server log file of 2,300,000 users' file-sharing behaviors in a P2P system. The results confirm the human dynamic model at the aggregate-level both in webpage browsing and P2P usage behavior. We have also found that there is detectable variability across the individuals in the decaying rate (i.e., the exponent gamma) of the power-law distribution, which follows well-known distributions (i.e., Gaussian, Weibull, and log-normal).
Computers and Society,Social and Information Networks