Community data portraiture : perceiving events, people, & ideas within a research community

Abstract

As a research community grows, it is becoming increasingly difficult to understand its dynamics, its history, and the varying perspectives with which that history is interpreted and remembered. This thesis focuses on three major components of research communities: events, people, and ideas. Within each of those components exploring how to construct and answer questions to improve connectivity and elucidate relationships for community members. Assuming the artifacts of a community (its publications, projects, etc) model a representation of its nature, we apply a variety of visualization and natural language processing techniques to those artifacts to produce a community data portrait. The goal of said portrait is to provide a compressed representation viable for consumption by a new researcher to learn about the community they are entering, or for a current member to reflect on the community's behavior and help construct future goals. Rather than evaluating a general technique, the tools and methods were developed specifically for the MIT Media Lab community, general principles can then be abstracted from this initial practical application.