Probability models for user session analysis in a visualization system: Using previous sessions advantageously |
Posted on:2007-01-27 | Degree:Sc.D | Type:Dissertation |
University:University of Massachusetts Lowell | Candidate:Chiang, Chih-hung | Full Text:PDF |
GTID:1448390005965338 | Subject:Computer Science |
Abstract/Summary: | |
Visual data exploration is an iterative discovery process that involves continual interactions between the user and computer. The goal of this activity is to gain knowledge about the data. However, observing the discovery process itself can also provide valuable information about the data, the user and the system.; Previous research into the visual exploration process has tended to focus on a single session, and often provided limited or no probabilistic analysis. This research investigates methods for analyzing multiple user sessions. Its two major components are probabilistic modeling based upon statistical analysis of multiple sessions and techniques for visualizing multiple user sessions. Combining visualizations of the current and previous sessions with probabilistic modeling provides tools to improve users' orientation in the current session, and in a broader context of actions they and others have taken in similar situations during past sessions that represent possible navigation choices.; The primary goal of this research is to provide facilities to visually explore and statistically understand how users interact with a visualization system during discovery sessions. This information may be used both to provide assistance for users or groups of users, and to verify and enhance the design of visualization systems. |
Keywords/Search Tags: | User, Sessions, Visualization, System, Previous |
|
Related items |