Statistical Physics in the Era of Big Data.

摘要

With the wealth of data provided by a wide range of high-throughout measurement tools and technologies, statistical physics of complex systems is entering a new phase, impacting in a meaningful fashion a wide range of fields, from cell biology to computer science to economics. In this dissertation, by applying tools and techniques developed in statistical physics, I present some of my contributions to the emerging field of Big Data in three distinct but related settings. First, we investigate long-term predictability of scientific impact. By deriving a mechanistic model for the citation dynamics of individual papers, we demonstrate that citation histories of all papers follow the same universal temporal pattern, helping us uncover the basic mechanisms that govern scientific impact. Second, we study the contextual factors that affect information spreading processes. We find that the social and organizational context significantly impacts to whom and how fast people forward information. Yet the structures within spreading processes can be well captured by a simple stochastic model, indicating surprising independence of context. Lastly, we study the mobility patterns and social interactions of mobile phone users, demonstrating the possibility of using the similarities between individual trajectories to predict social ties.