Date of Award


Degree Type


Degree Name

Doctor of Philosophy (PhD)


Computer Science

First Advisor

Ying Zhu - Chair

Second Advisor

Raheem A. Beyah

Third Advisor

Yichuan Zhao

Fourth Advisor

G. Scott Owen


Recent years have seen a growing interest in the emerging area of computer security visualization which is about developing visualization methods to help solve computer security problems. In this thesis, we will first present a method for measuring the complexity of information visualization designs. The complexity is measured in terms of visual integration, number of separable dimensions for each visual unit, the complexity of interpreting the visual attributes, number of visual units, and the efficiency of visual search. This method is designed to better assist fellow developers to quickly evaluate multiple design choices, potentially enables computer to automatically measure the complexity of visualization data. We will also analyze the design space of network security visualization. Our main contribution is a new taxonomy that consists of three dimensions – data, visualizations, and tasks. Each dimension is further divided into hierarchical layers, and for each layer we have identified key parameters for making major design choices. This new taxonomy provides a comprehensive framework that can guide network security visualization developers to systematically explore the design space and make informed design decisions. It can also help developers or users systematically evaluate existing network security visualization techniques and systems. Finally it helps developers identify gaps in the design space and create new techniques. Taxonomy showed that most of the existing computer security visualization programs are data centered. However, some studies have shown that task centered visualization is perhaps more effective. To test this hypothesis, we propose a task centered visualization design framework, in which tasks are explicitly identified and organized and visualizations are constructed for specific tasks and their related data parameters. The center piece of this framework is a task tree which dynamically links the raw data with automatically generated visualization. The task tree serves as a high level interaction technique that allows users to conduct problem solving naturally at the task level, while still giving end users flexible control over the visualization construction. This work is currently being extended by building a prototype visualization system based on a Task-centered Visualization Design Architecture.