The impact of time-based text visualization on situation awareness

Abstract

Overabundance of information has been well documented as a drag on human performance. Theoretical constructs such as visual dominance have led to visualization systems as a palliative, integrating cumbersome amounts of incoming data into more tractable forms. Most such systems have accepted only numerical inputs. I designed, developed, and evaluated a time-based visualization system for textual information, using situation awareness theory to evaluate the system. The system analyzes texts, such as emails and instant messages, as they arrive on the desktop and presents a labeled display, using suffix tree clustering and multidimensional scaling, of similarity between texts, updated as the mix of incoming texts changes. The system aids decision makers subject to information overload. I evaluated the system's performance under a simulated crisis in which groups of 30 students played the role of advisors to the University president during a recreation of an actual campus shooting. Half of the students used the visualization system to receive input, while half used a competing text-based system. Subjects were evaluated using an instrument drawn from situation awareness theory and found to achieve higher levels of situation awareness using the visualization system. Additional measures collected provide some guidelines for design of text-oriented, time-based visualization systems.