(reading of the borkin paper – interesting to look at last year’s Mizbee discussion notes)
There is a science to this: by using the ideas we’re going through this semester, you can come up with solutions that are different, and maybe better. (for some definition of science, this, and better)
Different Threads: (can they be seperated?)
- The style of paper (as a vis paper) – Case Study / Evaluation
- The work of their evaluation
- The design of their system
- Is it a good paper?
- Is it good visualization research?
- Is it good visualization practice?
- Is it helpful to the domain scientist?
Questions about a paper itself
- What is the venue?
- Who was the audience?
- What are they trying to “do” with the paper?
One thing that caught my attention was that all three readings have a figure before the abstract, which is different from CS papers that I have read
My Thoughts
- Case Study Paper / Eval Paper
- Specific to a specific problem – but what can we learn for our problems
- Focused on evaluation (not just for academic reasons – they needed to convince their users)
While reading the paper, however, I found that the content differed in a number of ways from a traditional paper that might describe a similar system.
Vis Ideas really come through:
- Process
- Color (and encodings)
- Mapping (spatial)
- Standard solutions worth questioning
- Empiricism vs. Processism
Although seemingly obvious, I thought it was notable that they broke this task in to simple categories like projections, dimensionality (2d vs 3d), color(s), and layout.
What I found interesting in this paper was that while 3D may make the most sense as a way to visualize information that is technically 3D in nature, this is not necessarily the most efficient way for people to take in that information
I cannot understand is what is preventing the process from becoming more simplified? There is no need for a human to make a subjective decision (the WHY VIS question)
It just reinforced the idea that a good visualization can remove/hide some of the information contained in the data it is visualizing to more effectively explore/explain that data.
Process
- qualitative study / requirements analysis
- iterative design
- use vis ideas so they didn’t have to search (3D, color maps)
- quantitative experiment and design
Experiment
- is 21 a lot or a little?
- quantitative nature – good/bad
- technical issues in design (latin squares, counterbalance, …)
- pretty compelling results
- CS folks are not great experimentalists – but we’re getting better
Not surprising
- Good color maps are good
- Good mappings (special purpose design) is good
- User centered process is good
Other great Points:
- user’s lack of self-understanding
I’m curious if the 3D model would be the more effective form for presenting/explaining information to patients and laypersons.
Their choices didn’t seem to correspond to any particular elements of color theory. Are there combinations they didn’t include that would work even better than red to black?