| As we enter the age of big and messy data, visualizations have emerged as an integral part of data analysis and decision making. Visualizations serve as an extension to the analyst's cognition, and the computer supports the human by managing large amounts of data, performing computationally difficult tasks, and providing relevant feedback to the human. As a result, analysts are now able to solve historically challenging problems in many areas including health, business and military.;Unfortunately, no two humans are the same, and a single human's behavior may even change from time to time due to fatigue, emotional and situational changes. Despite the overwhelming evidence of the impact of individual differences, most systems are built using a one-size-fits-all approach. To better support the analysis, the computer must be able to adapt and adjust to its user, and intelligently tailor situationally appropriate information to the user in a timely manner. Such systems represent the next generation of user interfaces in which the user is not forced to adapt to the system's interface and design, but is free to interact with the system in ways that are natural and in tune with her individual needs. This dissertation works toward creating such next generation visualization systems by exploring: (1) how individual differences such as personality affect the ways people use visualizations to solve problems, (2) how we can automatically detect these traits from a user's interaction with visualization tools, and (3) how we can leverage users' individual cognitive differences in a mixed-initiative system. |
