Research

Visual Analytics: Human Computation

Coming soon!

Visual Analytics: Visualizing Agent-Based Simulation Data using Graph Theoretic Approaches

Hairball in Color

Research in social and behavioral science relies heavily on simulation methods for modeling complex dynamic problems such as political violence, social destabilization, crime, and more. At present, social and behavioral scientists have access to robust agent-based simulation tools for modeling these issues. These models can contain thousands of agents, generating gigabytes of data representing the behaviors and states of the agents over thousands of simulations. However, these scientists have limited means to for analyzing the results generated by these thousands of simulations.

Zoom: Twins Zoom: Hairball Zoom: Hairball
Failed Stabilization

This research proposes a novel graph-theoretic approach to exploring these generated datasets. By visualizing the data as a temporal graph, where each node represents a (characterization of) a configuration of the agents and a directed edge represents a single-timestep transition from one configuration to another, we hope to better understand the simulation space. This model also allows us to leverage the significant body of existing work in graph analysis and feature detection, which we can use to compute, recognize, and direct our attention to interesting areas of the simulation space that might otherwise have been overlooked.

Developmental Technologies: CHERP

CHERP Logo

CHERP (Creative Hybrid Environment for Robotic Programming) is a hybrid tangible/graphical computer language designed to provide an engaging introduction to computer programming for children in both formal and informal educational settings. With CHERP you can create programs for robots like the LEGO Mindstorms RCX. It is currently being used for the Tangible Kindergarten project at the Tufts University Developmental Technologies Group.