Advertise here.

VisWeek2.jpg

VisWeek is a venue that joins 2 conferences and 1 symposium on the topics of Information Visualization, Scientific Visualization and Visual Analytics. This year just over 800 people met in Atlantic City, USA to discuss, present, or listen to recent research on all kinds of visualization related topics. If you want to skip our summary and see the "raw data", check out the videos captured of workshops, tutorials, panels, and paper talks. For a summary of our - of course highly subjective - experience, please read on.

MORE

VisWeek 2009

The week started on Sunday with workshops and tutorials. One co-author of this post co-organized a workshop (CoVis) on collaborative visualization for interactive surfaces such as wall and tabletop displays. About 90 people attended the workshop (we had expected about 20!) and we hope to see a lot more vis applications for these environments in the next couple of years.  Of course, there was also the popular VisWeb Workshop (previously advertised here) (editor's note: infosthetics, although not present, was also one of the organizers) - see Robert Kosara's description on his blog for more details.

IEEE Symposium on Visual Analytics Science and Technology (IEEE VAST)

On Monday the VAST symposium started. Here a short overview of our highlights:

The importance of storytelling was the topic of the VAST keynote by Andrew Glassner (well known from books such as Graphics Gems). It has been interesting to think about applying the power of storytelling - as we know it from books - to the presentation of data. He also presented his new book on interactive storytelling and more references on the topic can be found there.

The VAST conference this year included a number of papers on text mining and text analysis. Here, we would like to highlight one paper called "Parallel Tag Clouds" (YouTube video), or PTC (Collins et al.). PTC was built to visualize the contents of about 600,000 court decisions. It shows how case topics differ from place to place across the US by displaying linear vertical tag clouds of the cases and by highlighting commonalities across court districts (as known from parallel coordinates). The whole system is interactive and you can explore detail-on-demand all the way down to the individual court decisions.

Vis09_ParallelTagClouds.png


In the VAST closing panel Sarah Cohen, Lawrence Hunter and Joe Parry talked about data analysis in their respective fields: what data characteristics people deal with and how data analysis is commonly performed. It was quite inspiring to think about the challenges for visualization in these different domains. Sarah talked about her experience as an investigative journalist and the challenges arising from the multitude of different data sets that are often imperfect (because of OCR errors), need to be looked at in various ways, with no clear goals, typically under serious time pressure. Hunter posted a different picture for the area of biology, where researchers take considerable time to analyze huge amounts of data that is typically well-structured and of the same kind. Joe Parry talked about the role of interactive technologies and side-by-side paper printouts in the intelligence domain. On his blog you can even read a recap of his talk.

InfoVis + Vis:

After the VAST conference was over, both the InfoVis and Vis conference started in parallel to each other. The co-authors of this post are from the InfoVis community and, hence, you will see more projects from the InfoVis conference presented below. Nevertheless, the best paper award winner (Everts et al.) from the Vis conference certainly deserves a reference. Below you see a screenshot of a dense line dataset from the brain visualized with depth-dependent halos around the black lines. Each black line represents a (possible) fiber tract in the brain. The halo technique here gives a really good impression of fiber bundles and fiber distances and it is also well suited for print (black and white only!) and 3D viewing. Check out the authors' website and Youtube video for both 2D and 3D videos of the technique.

Vis09_Halos.png

Keynote by Colin Ware

The Vis/InfoVis keynote was held by Colin Ware, a visualization veteran and author of Information Visualization: Perception for Design. He started out with the observation that there is a gap between more theoretical models such as Anderson's ACT-R and more design-oriented "implicit wisdom" such as Shneiderman's visual information-seeking mantra. In the main part of his talk, Ware demonstrated several visualization tools for analyzing social network analysis and whale movements. Besides giving several heuristics, e.g., turning time into spatial patterns, Ware pled for a better integration of perceptual and cognitive rules into the design practice. Talking about the affordances of visual perception for cognition Ware quoted O'Regan with "the world is its own memory".

Interesting Papers from the InfoVis Conference

At VisWeek, we saw some interesting work that is also freely available for everyone to try and use. The first we'd like to point out is MizBee (Meyer et al.). It was written to support biologists with comparative genomics tasks. The paper was highlighted by an honorable mention at the InfoVis conference. If you're interested or working with biologists you may want to check this work out.

Vis09_Mizbee.png

Code_swarm (Ogawa and Ma) is a more playful example of visualization for conveying development activity in the open source community using an interactive particle system. It was previously presented on infosthetics, so refer to this post for more information.

ProtoVis (Bostock and Heer), is a jQuery-like library for developing interactive visualization in the browser using Javascript and SVG. Check it out if you are interesting in putting up your own data in a custom graphical manner online.
Vis09_ProtoVis.png

ABySS-Explorer (Nielsen et al.) is a visualization software for genome sequence exploration. Unfortunately there is no website yet for the project but the picture below give you an idea of what an overview of a genome sequence assembly can look like. The devil is in the detail as the squiggles, round connections between the lines, or the color amongst other details encode specific aspects of the sequence. This paper won one of the two best paper awards at InfoVis this year and is well worth reading in detail if you are interested in BioVis.
vis09_abyss.jpg

Judging from several presentations, a dedicated workshop on this topic, and conversations there is a steady move of visualization research and practice towards the web. For example, Matt McKeon's "Dashiki" (or Many Eyes Wikified) allows wiki-based authoring of visualization dashboards. Wiki pages can on the one hand include referenced or raw data and on the other hand multiple visualizations of this data. These pages can then also be embedded on external sites.
Vis09_Dashiki.png

Two additional extensions to IBM's Many Eyes that were presented at VisWeek were previously featured on infosthetics: Phrase Nets by van Ham et al. (infosthetics entry) and Wordle by Viégas et al. (infosthetics entry).

Evaluation was the focus of two papers we would like to mention. A Nested Model for Visualization Design and Validation (Tamara Munzer) talked about how to choose evaluation methods for information visualization on four intertwined levels as seen in the picture below. If you are interested in evaluating your tools or techniques this may be an interesting starting point for you.

Vis09_NestedModel.png

A second paper on evaluation (Bresciani and Eppler) discussed the use of visualizations in collaborative knowledge-intensive work such as the discussion of business strategies. The researchers conducted a year-long study with over 100 managers (in groups of 5) and found that groups who kept track of their conversations/discussions using visual representations were objectively more productive, had higher knowledge gain and better quality results than those goups that used only flipcharts. Interestingly, however, when groups were asked to subjectively rate their experience and results there was no difference between groups using visual representations or only flipcharts. So, maybe, sometimes using visualizations can make you perform better without you realizing it? Well, it's definitely something to think about...


This concludes our very very short summary of VisWeek. If you were at VisWeek, what were your favorites and why? What did you miss at this year's VisWeek?

Of course, there were many more interesting papers at the conference so please check out these external links to some papers, videos, and slides. 

Additional information:


This post was co-written by Petra Isenberg and Marian Dörk from the University of Calgary, Innovations in Visualization Laboratory

4 COMMENTS

Don't forget VizSec was also co-located (and at the same time as CoVis). http://blog.eqnets.com/2009/10/12/vizsec09/ has a brief summary of a few papers there.

Fri 23 Oct 2009 at 12:36 AM

Yes! I would encourage everyone to take a closer look at the program to see what else went on at the conferences. We didn't write about anything that we did not attend (like VizSec) but that does not mean that there wasn't exciting stuff going on there. We left a lot of exciting stuff out unfortunately :-(.

Also worth checking out are the twitter streams: http://twitter.com/#search?q=%23visweek

Fri 23 Oct 2009 at 1:06 AM

Great writeup! VizWorld covered the event as well ( at http://www.vizworld.com/tag/visweek ) although we focused more on IEEE Vis and the Tutorial sessions.

Fri 23 Oct 2009 at 2:01 AM

Neat stuff, thanks for the recap. Any idea if/when Infovis or VAST session video is going to be posted? I had to skip most of the Graph Visualization session (and all of VAST!) and would love to see what I missed.

Sat 24 Oct 2009 at 2:37 PM
ADD A COMMENT
Commenting has been temporarily disabled.