2016 Graduate Workshop in Complexity and Computational Social Science

Student Projects

Each student began a research project during the two-week workshop. Below are brief descriptions of these various projects. These projects will form the basis for dissertation chapters and journal articles.

Thomas Briggs, Computational Social Science, George Mason U. (tbriggs@gmu.edu).

Tom is creating "emotional" agents by incorporating key aspects of the "big 5" personality factors (openness to experience, conscientiousness, extraversion, agreeableness, and neuroticism) into each agent's behavior. In his initial work, he considers a model of team behavior with employees driven by conscientiousness and extraversion, and finds that team performance can be linked to key interactions across these traits. He is currently validating the above results using available data, and extending the modeling approach to include the other three personality factors.

Julia Eberlen, Social and Cultural Psychology, Université Libre de Bruxelles (jc.eberlen@gmail.com).

Julia is intrigued by the learning of stereotypes---shared beliefs about a group---in networks. To model this process, she uses ideas from statistical learning to allow her Bayesian agents to to influence one another based on their network connections. This model allows her to make some key connections between the processes of opinion formation and stereotyping.

Claudius Grabner, Economics, U. of Bremen (graebnerc@uni-bremen.de).

Claudius is exploring the emergence of institutions using evolving automata to capture strategic adaptation. In each period, strategies interact with a small group of the other strategies given a set of connection probabilities. When the connections are fixed across periods, the strategy chosen by the more central players is the one that most influences global behavior. When the network can evolve over time, a very different set of strategies emerges, resulting in quite different outcomes from the fixed-network world. Moreover, the evolving networks take on characteristic, asymmetric forms.

Merritt Hughes, Public Policy, U. of Massachusetts Boston (Merritt.Hughes001@umb.edu).

Merritt is interested in building a model that can evaluate detailed linkages between emission permit trading and capital investment in electricity generation. As a first step she created linked auctions for electricity supply and permit demand and found that the initial agent technology mix significantly influenced p rice and profit paths toward equilibrium. Using such a model, she will next explore the interaction among policy choices, sustainable production technology, and consistent supply.

Robert Moulder, Quantitative Psychology, U. of Virginia (rgm4fd@virginia.edu).

Bobby is creating models where an agent's behavior is determined by combining multiple models drawn from disparate areas of quantitative psychology. His current model combines existing theories tied to cognitive regulation, attitude polarization, and social comparison. Such combinations result in novel behaviors, such as bifurcation points that can result in polarization depending on the heterogeneity of the agents. These results can be used to explore information sharing across groups with diverse individuals.

Antonio Sirianni, Sociology, Cornell (adsirianni@gmail.com).

Tony is analyzing hierarchy formation in social systems with conflict. In his system, pairs of agents may establish a dominance ranking when they engage in conflict. Other agents may observe, and remember, the outcomes of such conflicts. He finds that as the agents' memories of past outcomes increase, more hierarchical structures emerge. He also identifies an important tradeoff between social harmony and adaptation.

Martin Smyth, Technology and Society, Stony Brook (martin.smyth@stonybrook.edu).

Marty is developing models of regime violence during political revolutions. His focus is on the interplay between protesters and the response of the regime in power. His model includes factors such as political grievance, activation accretion, information flows, and the ability of the regime to project force. Using this model and data that arose in Egypt during the Arab Spring, he is formulating a more general theory of political revolutions.

Dandong Yin, Geography and Geographic Information, U. of Illinois at Urbana-Champaign (ddcamiu@gmail.com).

Dandong is using twitter data to understand social patterns across time and space. Clustering th data based on each tweet's location, he can identify various social patterns, such as daily activities linked to the time of day. These patterns reveal the presence of public versus private places, including specific locations like restaurants or popular outdoor meeting places. The data may also reveal other interesting social behaviors, such as shirking at work.

Luis Alberto Sanchez Zacateco, Economics, Instituto Politecnico Nacional (luisal.zac@gmail.com).

Luis considers a multi-objective evolutionary approach to understanding stock-market behavior. His model considers two representative agents, one of which invests using a traditional Markowitz model while the other uses a genetic algorithm to maximize a measure of the Conditional Value at Risk (CVaR). Using data from the Mexican Stock Market over a three-year period, he finds (confirmed by back testing) that the evolving algorithm can develop productive trading strategies.

Xiaolin Zhuo, Sociology and Quantitative Social Science, Harvard (xiaolinzhuo@fas.harvard.edu).

Xiaolin is doing a network analysis of research on the topic of networks---a rapidly growing transdisciplinary research field. Her analysis considers the various overlapping networks that are common in research, such as coauthorship and citation patterns. She finds that the differnt networks have very different characteristics, and she is currently building a theoretical models to understand better the forces that can lead to such observations.

John H. Miller , miller@santafe.edu.