Busy New York scene
Network Systems Science and Advanced Computing | Resilient Societies and Interdependent Infrastructures | Social and Behavioral Network Science | Social, Cognitive, and Behavioral Science

Leveraging Sociological, Economic and Psychological Theory for Simulating At-Scale Online Information Propagation

Contacts
Sponsor

DARPA

The goal of DARPA’s Computational Simulation of Online Social Behavior (SocialSim) is to develop innovative technologies for high-fidelity computational simulation of online social behavior.

A simulation of the spread and evolution of online information, if accurate and at-scale, could enable a deeper and more quantitative understanding of use of the global information environment than is currently possible using existing approaches. High-fidelity (i.e., accurate, at-scale) computational simulation of the spread and evolution of online information would support efforts to analyze strategic disinformation campaigns, deliver critical information to local populations during disaster relief operations, and could potentially contribute to other critical missions in the online information domain.

Project Overview

The simulation of online information propagation at-scale provides a method for developing scientific understanding of several critical phenomena in sociology, psychology, economics, political science and other human-related systems. The integration across disciplines and modeling approaches, and across scales in some cases, has not yet matured into a set of standard methods nor has a set of example computational platforms become available for general use in government, academia or industry.

To tackle this problem we have developed an agent-based modeling platform for the implementation of a set of methods that provide

  1. development and testing of social, economic and psychological theory at-scale, and
  2. providing predictive (in time) forecasting of information propagation, and
  3. providing situational forecasting (what-if scenario simulation).
Findings

Key Findings:

Development of a paradigmatic view of social simulation based on a review of the state-of-the-art in modeling human systems.

Human social group behavior modeled as extended cognition makes sense.

Modeling agents with high-fidelity cognitive architectures is feasible at scale.

Development of a methodology for systematically integrating sociological, cognitive and economic functions into a nested functional Bayesian agent.

Development of a new semantic embedding method for social media platforms that is designed for interfacing with psychological models.

Development of a semantic clustering method.

Social psychological constructs implemented as simple computational agents can be used to predict future user belief production on social media.

Sociological constructs implemented in agent-based representation of social media system predicts the evolution of network structure.

various charts
Figure 1. An illustration of sociologically driven network analysis of users on a social media platform (Twitter) that was used to drive the training of a psychological model of attitude production for simulation of agent-level behavior in the Matrix agent-based simulation platform.  The method used a sociological construct called “factions” to cross-classify communities by semantic similarity called narrative elements or narratives (see Panels A and B; each node in the network is a community).  Panel A shows six factions within a large Twitter network.  Panel B illustrates the core narrative position of the second faction (green nodes in the network). Community structure is used to differentially train a psychological model of attitude production (shown in Panels C and D).  Panel C shows the structure of the attitude production model.  Panel D shows the performance of the attitude production model with respect to its expected production of narrative elements given very structured and sparse inputs.  A key feature of the attitude production model is behavior that reflects a compression of the input to output mapping.
Tools and Publications

Tools

Matrix ABM Simulator

Books Articles and Proceedings

Afrasiabi, M., Orr, M.G., and Austerweil, J.L.; “Evaluating Theories of Collaborative Cognition Using the Hawkes Process and a Large Naturalistic Data Set”; Paper presented at the Annual Meeting of the Cognitive Science Society; July 2019

Swarup, S.; “Adequacy: What Makes a Simulation Good Enough?”; Paper presented at the Spring Simulation Conference; May 2019

Orr, M.G.; “Multi-Scale Resolution of Human Social Systems: A Synergistic Paradigm for Simulating Minds and Society”; Social-Behavioral Modeling for Complex Systems; edited by Davis, P.K., O’Mahony, A., and Pfautz, J.; John Wiley & Sons, 2019, pp. 697-710.

Swarup, S., Marathe, A., Marathe, M.V., and Barrett, C.L.; “Simulation Analytics for Social and Behavioral Modeling”; Social-Behavioral Modeling for Complex Systems; edited by Davis, P.K., O’Mahony, A., and Pfautz, J.; John Wiley & Sons, 2019, pp. 617-632.

Orr, M.G., Lebiere, C., Stocco, A., Pirolli, P., Pires, B., and Kennedy, W.G.; “Multi-scale resolution of neural, cognitive and social systems”; Computational and Mathematical Organization Theory, 25 (2019): 4-23.

Orr, M.G., Lebiere, C., Stocco, A., Pirolli, P., Pires, B., and Kennedy, W.G.; “Multi-Scale Resolution of Cognitive Architectures: A Paradigm for Simulating Minds and Society”; Paper presented at the Association for the Advancement of Artificial Intelligence (AAAI) Fall Symposium; October 2018

Bhattacharya, P.; “The Matrix: An Agent-Based Modeling Framework for Complex Systems and Data Intensive Simulations”; Paper presented at the Dynamics On and Of Complex Networks Workshop (DOOCN-XI) at the Complex Systems Conference (Thessaloniki, Greece); September 2018

Orr, M.G., Lebiere, C., Stocco, A., Pirolli, P., Pires, B., and Kennedy, W.G.; “Multi-scale Resolution of Cognitive Architectures: A Paradigm for Simulating Minds and Society”; Social, Cultural, and Behavioral Modeling; edited by Thomson, R., Dancy, C., Hyder, A., and Bisgin, H.; Springer, 2018, pp. 3-15. [Presented at the International Conference on Social Computing, Behavioral-Cultural Modeling and Prediction and Behavior Representation in Modeling and Simulation (SBP-BRiMS); July 2018]

Conference Presentations

Bhattacharya, P., Ekanayake, S., Kuhlman, C.J., Lebiere, C., Morrison, D., Swarup, S., Wilson, M.L., and Orr, M.G.; “The Matrix: An Agent-Based Modeling Framework for Data Intensive Simulations”; Talk presented at the International Conference on Autonomous Agents and Multiagent Systems (AAMAS); May 2019

Orr, M.G.; Panel presentation on Miscellaneous Topics; Invited panel presentation for “Physics of Information Processing” at the 3rd International Workshop on Social Sensing (Orlando, Florida); April 2018

Awards

Best Paper for the International Conference on Social Computing, Behavioral-Cultural Modeling and Prediction and Behavior Representation in Modeling and Simulation (SBP-BRiMS) 2018

Team

Research Associate Professor

Research Scientist

Research Scientist

Research Associate Professor

Research Associate Professor

Postdoctoral Research Associate

Research Associate Professor

Senior Project Manager

Research Program Advisor

In the News

Lebiere, C.; “CMU researchers are building a model to predict human behavior. It could save lives one day.”; Interviewed by Linder, C.; Pittsburgh Post-Gazette; February 6, 2018