ICES: Small: Discovering Fundamental Structural and Behavioral Laws of Social Networks

For centuries, the discovery of scientific laws has been driven by empirical investigation of natural phenomena. However, human capabilities of relationship discovery get strained in multi-dimensional spaces, which in particular may explain why we have not seen any fundamental laws postulated for the structure or behavior of such complex constructs as social networks. This project will exploit the methodological toolbox based on symbolic regression with the objective to assist the human mind in simultaneously searching through multiple social network datasets and detecting stable relationship patterns across them, thus allowing for the discovery of natural social network behavior laws. This project will identify objectively measurable, macroscopic metrics pertaining to social network phenomena and explore under what conditions the computer-enabled empirical approach to knowledge discovery can be successful. The scientific inquiry approach based on the computer-enabled empirical learning paradigm will rely on recent advances in stochastic processes, systems theory, symbolic regression, evolutionary algorithms, and data fusion. Using standard social network metrics, computational experiments will be conducted and their results manually traversed in search for robust expressions (explicit formulae and invariants) corresponding to hidden symbolic laws. This project will benefit organizations and institutions that rely on fundamental understanding of social structures. It will further advances in communications, economics, engineering, psychology, sociology, epidemiology, and other domains engaged in studies of social interaction and behavior. On a broader level, this project will enrich the methodological toolbox of inquiry for other unexplored areas of modern science. Through student involvement activities and outreach, this research effort will enhance the educational environment supported by NSF and fostered at the University at Buffalo.