Hannes Pretorius

| About . Research . Publications

Visualization of State Transition Graphs

For my PhD research, I investigated ways to visualize large state transition graphs. State transition graphs (also called state transition systems, state spaces or finite automata) play an important role in computer science and engineering: they are used to describe the behavior of computer-based systems whose states evolve over time.

Conceptually, state transition graphs are straightforward. In these graphs, every node represents a state a system can be in and every arc represents a transition from one state to another. However, for modern computer-based systems they tend to be quite large. They often contain tens of thousands of nodes, or more. As a result, their complexity increases and it becomes difficult to study the systems they model.

To address this issue, I developed a number of techniques for the interactive visualization of state transition graphs. Much of the work takes advantage of multivariate data associated with nodes and edges. Using an experimental approach, several new methods were developed in close collaboration with target users.

The results of my research are described in my PhD dissertation:

Visualization of State Transition Graphs
A.J. Pretorius
PhD dissertation, Eindhoven University of Technology, ISBN 978-90-386-1405-2, 2008.

An article about this research appeared in the Belgo-Dutch technology press:

Visualisatietools brengen toestandsdiagram tot leven
Bits&Chips, November 2008.

We took a retrospective view on the process of designing visualization tools for domain experts in an article for the Information Visualization Journal:

What does the user want to see? What do the data want to be?
A.J. Pretorius and J.J. van Wijk
To appear in Information Visualization Journal, vol. 8, no. 3, 2009.

Multivariate graph visualization

Most graph visualization techniques focus on showing the structure of the graph and do not offer much support for dealing with node attributes and edge labels. We developed a technique where this type of data plays a more central role. This enables users to detect relations and patterns based on data associated with nodes and edges. We built a prototype that implements this technique, which is available here.

We described our approach and design methodology in a paper for EuroVis 2008:

Visual inspection of multivariate graphs
A.J. Pretorius and J.J. van Wijk
Computer Grahics Forum (Proc. Eurographics/IEEE EuroVis), vol. 27, no. 3, pp. 967-974, 2008.

Trace visualization

Our method combines three perspectives that can be taken on multivariate trace data: (1) as a function of attribute values assumed, (2) as a function of time and (3) as a state transition graph. The combination of all three perspectives provides the user with a rich visual interface that enables gaining insight into system behavior. You can learn about our prototype here.

We described our approach in a paper for Pacific Vis 2008:

Multiple views on system traces
A.J. Pretorius and J.J. van Wijk
Proc. 2008 IEEE Pacific Visualization Symposium, Kyoto, Japan, pp. 95-102, 2008.

User-defined diagrams

DiaGraphica enables users to define and parameterize custom diagrams that closely reflect associated semantics. User-defined diagrams are incorporated in a number of correlated visualizations for studying multivariate graphs. Learn more about DiaGraphica.

We described our technique in an IEEE CG&A article:

Bridging the semantic gap: visualization of transition graphs with user-defined diagrams
A.J. Pretorius and J.J. van Wijk
IEEE Computer Graphics and Applications, vol. 27, no. 5, pp. 58-66, 2007.

Attribute-based visualization

NoodleView is a prototype for the interactive visualization of multivariate state transition graphs. It offers interactive attribute-based clustering and a number of other interactive features. Clustering reduces the size and complexity of these graphs and is also a useful analysis tool. Learn more about NoodleView.

We described our approach in a paper for InfoVis 2006:

Visual analysis of multivariate state transition graphs
A.J. Pretorius and J.J. van Wijk
IEEE Transactions on Visualization and Computer Graphics, vol. 12, no. 5, pp. 685-692, 2006.

Systems analysis

To experience first-hand what my target users do on a daily basis, I participated in the verified design of a real-world system. The focus was on the specification and verification of desired behavior of an automated parking garage. We also implemented a simple custom visualization plug-in that made our lives a lot easier.

We described our experience in:

Verified design of an automated parking garage
A. Mathijssen and A.J. Pretorius
Proc. 11th Int'l Workshop on Formal Methods for Industrial Critical Systems (FMICS 2006), LNCS 4346, pp. 165-180, 2007.

Multidimensional visualization

Transition systems pose an interesting multidimensional visualization challenge. Every state can be considered as a point in n-space. To experiment with this approach we built a prototype. StateVis provides a number of methods to project such "high-dimensional" states to 2D. Download StateVis.

Our approach is described in:

Multidimensional visualization of transition systems
A.J. Pretorius and J.J. van Wijk
Proc. 9th Int'l Conf. Information Visualisation (IV05), London, UK, pp. 323-328, 2005.

Copyright 2008. A.J. Pretorius. All rights reserved.