Interaction
From Visualization Sp06
Lecture on Feb 2, 2006
Readings
- Visual information seeking: Tight coupling of dynamic query filters with starfield displays, Ahlberg & Shneiderman. (html)
- Visual exploration of time-series data, Hochheiser & Schneiderman. (html) (pdf)
- Postmortem of an example, Bertin (handout)
- The visual design and control of the trellis display. Becker, Cleveland and Shyu. (ps)
Optional Readings
- Table lens, Rao and Card, (acm)
- Human guided search: Survey and recent results, Klau, Lesch, Marks & Mitzenmacher. (pdf)
- Design and evaluation of incremental data structures and algorithms for dynamic query interfaces. Tanin, Beigel & Schneiderman (citeseer)
Demonstrations
Contents |
Ryanaip - Feb 02, 2006 11:12:34 am
I've found the concept of timeboxes very useful in my work, in this case using linked visualizations to explore 2-dimensional spatial data. The example on the left uses spatial query boxes to highlight regions on a floor plan, and timelines are updated to show when tracked objects were in those regions. The example on the right uses timeboxes on the timeline to trigger the display of paths taken during that time period on the floor plan above.
Pushkar - Feb 02, 2006 11:53:46 am
I have a comment about 'TimeSearcher2' and its ability to display data trends over large periods of time. The concept of 'aggregation' is used to show the data trends when the number of samples is higher than the number of available pixels.
Firstly, I think aggregation should be utilized even if there are enough pixels to display all the samples - in my opinion a smooth plot can show data trends better than a noisy/spiky one. Secondly, more sophisticated filters (instead of simply averaging the aggregated data samples) that study the pattern of data will be much more helpful in selecting which samples to aggregate and how to aggregate them. For instance, if one is looking at a company's financial statements from 1970-present, there will always be unusual activity in April (end of the fiscal year) for every year. A smart aggregation scheme should be able to detect this and choose May-April as aggregation periods, so that we can get a better sense of how the company has done year-by-year.
Please correct me if the above functionality already exists in the TimeSearcher2 and I somehow missed it.
Cynbot - Feb 02, 2006 12:24:27 pm
In regards to the tight coupling paper, I also recommend reading the Apple Human User interface Guidelines [1]. It has some neat concepts regarding "forgivness" and "stability" that are similar to the points made in this paper. In regards to the TimeSearcher tool, I would have though that highlighting regions on the axes would be more intuative than drawing a box, esp when looking at craziness like Figure 1.
Bryan - Feb 02, 2006 01:38:01 pm
I really like the use of video zooming as a way to handle multiple scales of data and for exploring large or non-uniformly dense data sets. I think this is actually why Google's map tools are so much fun to use. My favorite example of this is something I saw about 10 years ago in an IMAX theatre - [Cosmic Voyage]. It's basically a continuous spatial zoom through about 43 orders of magnitude, and a blast to watch. You can get it on DVD now, it's worth a rent.
Maneesh - Feb 02, 2006 05:22:42 pm
Ryanaip - The example from your work looks really interesting. Do you also have the ability to draw boxes within the floorplans to select paths between particular rooms?
Bryan - Zooming is definitely important, and in fact we'll be talking about it in the next class quite a bit. I haven't seen Cosmic Voyage, but you reminded me of the Eames' Powers of 10 video that is a great example of zooming through many orders of magnitude.
Cynbot - Feb 03, 2006 04:55:24 pm
This may have already been mentioned but the University of Maryland's Piccolo tool does a nice job of zooming of many images. http://www.cs.umd.edu/hcil/jazz/
Lesliei - Feb 06, 2006 04:45:09 pm
In class, it was mentioned that users often complain that there are too many widgets on a visualization tool. However, having too many widgets is not necessarily a problem --- consider for example the cockpit of an airplane. Pilots have many instruments, but they can draw conclusions from them rapidly. How much effort can a designer assume users will expend learning a tool?
Nchentan - Feb 07, 2006 06:04:58 pm
Added to Lesliei's comment on the widgets: It seems that the design principle of the widgets in the visualization tool should probably be similar to the design principle of a good GUI. It seems that with proper grouping and great GUI layout and a bit of user training, having many widgets should not pose a problem.
