The Value of Visualization
From CS 294-10 Visualization Sp10
Lecture on April 26, 2010
- Visualization Research Challenges. NIH/NSF Research Report. Johnson et. al. (pdf).
- The Value of Visualization. Van Wijk. (pdf)
Jon Barron - Apr 26, 2010 07:37:29 pm
I'm very grateful the "virtual colonoscopy" paper was not required reason for this course.
A great quote: "[W]e should ultimately aim at generic results (models, laws) that enable us to understand what goes on and to predict why certain approaches do or don’t work."
Of everything we read in this class, the best works were always those which had *conclusions*, not just results. Things like "position is effective for encoding quantities" and "hue is effective for categories, but not quantities" or "shading is a weak cue for shape" are concise and actionable lessons that we can actually use, as opposed to "our visualization system worked well and is awesome, for myriad reasons which cannot be decoupled from one another".
Stephen Chu - Apr 28, 2010 03:11:29 am
Watching Hans Rosling's TED presentation made me realize how important and effective visualization can be in helping us understand public health issues. One of the visualization research challenges mentioned was the development of visualization techniques suited for understanding and dealing with health-related data. Although prevention is recognized as a more effective strategy than treatment, public health policies have been too slow in the processes of approval and implementation to address today's problems. Now, with more and more accessible data, visualization research is needed in order to help speed up policy advancements by allowing everyone to see the important messages in the data.
Arpad Kovacs - May 01, 2010 05:32:50 pm
Johnson: Visualization Research Challenges
This report is a review of the visualization research field, identifying several areas that could benefit from new or advanced visualization techniques. It also provides a general wish list and advocates government policy changes for improved support and funding of projects utilizing visualization. However, the report is very generic, and it does not provide many actionable items or solutions. Findings, such as "We must design appropriate interaction metaphors for both current and future hardware in order to harness the full power of visualization systems" and "A disproportionate percentage of the important visualization research is transitional" sound great, but what do they mean? The main value of the report probably lies in the overview of the different visualization application fields and the real life examples cited on the side bars.
Van Wijk: The Value of Visualization
This is an interesting article attempting to understand and quantitize the cost/benefit ratio of visualization. While judging the results of visualization is highly subjective, the proposed formula is an eye opener and helps to uncover the factors affecting the cost of visualization. I enjoyed the description of the sample applications, especially the problematic wave surface interpolation and the resulting insight: "interpolation is not by definition better than approximation, and the judgement of an expert is vital for proper interpretation and validation".
Zev Winkelman - May 02, 2010 11:42:33 pm
Van Wijk - I wish I had been told not to focus on the equations before I read the paper. But, nevertheless, an interesting challenge to the field. I thought the reference to "Many of us will have written paragraphs like the preceding one, where I attempted to give the standard rationale of our field" amusing, as I plan on having to write such a paragraph soon. Perhaps a boiler plate, GPL'd copy or something we could all just copy and paste ? I digress. The gap between research and users is great in many fields, so I'm not sure that calls for reconsidering the value of visualization are all that warranted. It could be the case that the power of the tools are so great that innovation is being conducted in proprietary, enterprise or other more closed environments. I can appreciate the perspective that Van Wijk offers that "in many cases just using a line graph is the best way to show a time varying signal", however, in the circumstances where a better method is appropriate and appropriately applied, the impact is still very high. I did think the last sentence was poetic: "aim for provable effectiveness and efficiency, aim for elegance and beauty, and aim at generic laws with predictive power."
Johnson et al. - Lots of goodies in this paper - "high bandwidth channel of human visual perception", "a wealth of information creates a poverty of attention", "suppose we knew everything", etc. This paper is a great overview of the state of the field, as it struggles for peer status with other fields. I thought the point about grid computing 'solving' the problems of speed and storage, but not of visualization to be very interesting. Looking forward to the future !
Jaeyoung Choi - May 03, 2010 02:10:55 am
van Wijk: I like this paper as it wraps up many interesting ideas, many of which comes from what I've learned from this class. I stongly agree with how high initial cost would hinder the use of new visualization technology as I would still need to spend substantial time to get familiar with both the tools and the visualization techniques and most importantly, to produce effective visualization. The claims from the advocates of "interaction is 'bad'" does have a point, since extraction of knowledge from data is NOT an objective process in practice, as mentioned in 4.4.
There could be trade-offs from a design choice and sometimes there is no explicit goalline to meet, which makes the evaluation of visualization to be tricky. Still, simple rule of thumb lessons that we learned from the class (like the ones that Jon mentioned above) will help us in a great deal.
DavidZats - May 03, 2010 02:56:09 pm
I really enjoyed the paper on the "the value of visualization." I think that trying to quantify which visualizations are/are not valuable is very important. While it is unlikely that we will ever settle on one specific equation for determining the value of visualization, this does bring up issues that people should consider when creating them. For example, I think the discussion about visualizations being used to prompt the viewer to take a decsion/action is very important. Additionally, I liked how the factors indicating the usefulness of a visualization include how many people it targets and how much information they gain.
Prahalika Reddy - May 12, 2010 08:49:27 am
The value of visualizations, I feel, is oftentimes very subjective. What one person sees in a visualization, the insights they gain into whatever is being depicted, is not always the same as what someone else sees. While there are many standard techniques that are effective for many visualizations, not all so-called "effective" or "ineffective" techniques are correct for every person. It's important to keep track of what the techniques are that are effective for most people so that at least visualizations are useful for a large number of people.
Visualization is one of the most important fields of interest. Without the ability to visualize massive amounts of data, it's not very useful to have that data at all. Therefore it's important to have the most effective visualizations. That's a very hard task and it makes effective visualizations that much more important and valuable.
Shimul Sachdeva - May 13, 2010 04:31:40 am
The paper brought up an important notion of "a wealth of information creating a poverty of attention, and a need to allocate that attention efficiently among the overabundance of information sources that might consume it". Although it is true in some aspects, it is still a vague statement since it is hard to quantify "too much information". The visualization discovery process that talks of the iterative process that making visualizations can be reminded me of Assignments 3 and 4. The treemap example in the paper was interesting too - although it was in some sense, a violation of the principle I mentioned above.