« When maps are usefull games... | Main | 2006 infovis competition ! »
August 15, 2006
An interview with...John Stasko
In the middle of this nice August, Im really glad to host the Johns words and thoguts about infovis.
John is a Professor Assoc. Chair, Division of Interactive and Intelligent Computing at the College of Computing at Georgia Tech.
1 - When, how and why have you decided to research on infovis ?
My doctoral thesis work was in the area of Software Visualization (softvis) which is a subdomain of infovis, so I was working in this area as early as the 1980s. For about 8 years after completing my degree and starting as a professor (1989), I worked primarily on softvis. Around that time I just became interested in more general issues with visualizing abstract data sets and I broadened the set of projects being worked on in my lab. I had always been interested in problems involving computer graphics and user interfaces, and I enjoyed playing with sports statistics as a kid. So infovis is kind of a natural fit for me.
When we (Georgia Tech) changed from quarters to semesters around 2000, we instituted a new set of courses.
I added a graduate course on infovis and teaching it has truly helped foster my interest in this area.
2 - Which is, to you, the most interesting project you have worked on and why ?
Its difficult to pick just one or two projects. For the last few years I have worked on a project called the InfoCanvas (http://www.cc.gatech.edu/gvu/ii/infoart) which is somewhat infovis-related, but not in the classic sense of an analysis system for large data sets. The InfoCanvas is a kind of electronic painting in which the objects in the scene change position, appearance, color or size to indicate the state of information that a person may care about. For instance, a bird in the sky may fly high when the stock market is up and low when it is down, or the number of apples on a window sill may indicate what tomorrows high temperature will roughly be. We put a picture frame around an LCD display to simulate an interactive painting. A key point is that the person running and using an InfoCanvas can choose what information is to be presented and how to present it.
Another project that I have enjoyed is a visualization technique called SunBurst (http://www.cc.gatech.edu/gvu/ii/sunburst) that uses a circular, space-filling depiction to represent hierarchical information structures. Im a big fan of treemaps as visualizations for this kind of data, and SunBurst provides an alternative depiction that I feel can be useful when one really wants to see the structure of the information hierarchy. We developed some interesting animated interactions to allow viewers to focus on particular pieces of the hierarchy.
3 - On what themes are you working now ?
Lately I have been involved in some more reflective projects in infovis that examine and question infovis systems utility for different types of tasks. I see far too many "solutions in search of problems in this area, visualization techniques that may be technical achievements but also seem to have little utility and application. I think we need to argue more rigorously on why visualization techniques and systems can be and are useful. An important step in doing this is to build a better understanding of the analytic and exploratory processes that people undertake. Where and how can infovis systems assist them?
Accompanying these efforts, I have been working on a project in the area of visual analytics, which is closely related to infovis.
Jim Thomas of Pacific NW National Labs in the U.S. has popularized this term and it involves an even closer examination of the analytic processes people undertake when assisted by visualization tools and systems.
4 - If you should suggest an education programme to a student interested in InfoVis, which University Course do you suggest him/her?
I think a strong interdisciplinary background would be ideal for students interested in this area. Seek out schools that offer such programs.
I would start with a strong base in computer science including computer graphics, human-computer interaction, and data bases or data management.
From there, courses in graphic design, art, and aesthetics would be invaluable. Furthermore, courses in experimental design and psychology can be helpful for assessing the utility and usability of systems. Perhaps most directly, a number of faculty who do research in infovis now offer courses on the topic, so students can learn about the area firsthand.
5 - How do you think infovis researches will evolve in your country for next years ?
I have a hunch that were beginning to be tapped out in terms of new and innovative visual representations. It is very difficult to come up with a new visualization that no one has seen before and that is truly useful. I think a lot of the innovation to come will be on the interaction side, how we make visualizations more interactive and more able to show multiple aspects of data.
I think it will be interesting to see if current infovis research techniques make their way into more commercial tools and systems.
We need to make the techniques very easy to use and understand in order for this to happen. On the other hand, I see the need for infovis systems that can handle massive, heterogeneous data sets. Here, the infovis systems may be so complex that users will have to undergo extensive training to learn how to use them, but once users are familiar, the system will enable much richer analysis and exploratory activities.
http://www.mentegrafica.it/blog/?p=119
Posted by InfoVis at August 15, 2006 6:10 PM
