COMPUTATIONAL SOCIAL SCIENCE

CSS SEMINAR - BARRETT

Friday, April 17, 3:00 p.m.
Center for Social Complexity Suite
Research Hall, Third Floor

Massively Interactive Systems: Thinking and Deciding in the Age of Big Data

Christopher L. Barrett
Executive Director, Virginia Bioinformatics Institute
Professor of Computer Science
Virginia Tech

ABSTRACT: This talk discusses advanced computationally assisted reasoning about large interaction-dominated systems. Current questions in science, from the biochemical foundations of life to the scale of the world economy, involve details of huge numbers and levels of intricate interactions. Subtle indirect causal connections and vastly extended definitions of system boundaries dominate the immediate future of scientific research. Beyond sheer numbers of details and interactions, the systems are variously layered and structured in ways perhaps best described as networks. Interactions include, and often co-create, these morphological and dynamical features, which can interact in their own right. Such “massively interacting” systems are characterized by, among other things, large amounts of data and branching behaviors. Although the amount of associated data is large, the systems do not even begin to explore their entire phase spaces. Their study is characterized by advanced computational methods. Major methodological revisions seem to be indicated.

Heretofore unavailable and rapidly growing basic source data and increasingly powerful computing resources drive complex system science toward unprecedented detail and scale. There is no obvious reason for this direction in science to change. The cost of acquiring data has historically dominated scientific costs and shaped the research environment in terms of approaches and even questions. In the several years, as the costs of social data, biological data and physical data have plummeted on a per-unit basis and as the volume of data is growing exponentially, the cost drivers for scientific research have clearly shifted from data generation to storage and analytical computation-based methods. The research environment is rapidly being reshaped by this change and, in particular, the social and bio–sciences are revolutionized by it. Moreover, the study of socially– and biologically–coupled systems (e.g., societal infrastructures and infectious disease public health policy analysis) is in flux as computation-based methods begin to greatly expand the scope of traditional problems in revolutionary ways.

How does this situation serve to guide the development of “information portal technology” for complex system science and for decision support? An example of an approach to detailed computational analysis of social and behavioral interaction with physical and infrastructure effects in the immediate aftermath of a devastating disaster will be described in this context.