Georges Grinstein at UMass just asked me what the top 5 research challenges are for the field of information visualization. I thought I'd blog my response here that I wrote up in 15min:
About 5-6 years ago I wrote up an article that was going to be published in a software engineering magazine about challenges in Visualization, and then they pulled the special theme after I submitted it. I wonder where that article is now....
In any case, briefly here is my updated list off the top of my head (in no particular order):
- Integration with data sources: This remains to be one of the major challenges that never seems to go away. It's just too damn hard most of the time to transform the data into a format that visualization tools can understand.
- Integration of interactivity with analytic algorithms: Most of the time visualization is still run with command lines in batch mode. it's still too hard to ask the 'what-if' questions. This was what my visualization spreadsheet system was trying to solve.
- Working together with social analysis of data sets: Most of the time visualization exists in a context where the data is being analyzed or consumed in a social setting. It's not create a pretty picture and then stop and done. It exists in a social setting where there is a lot of different annotations by different people and different ideas being tried. Think ManyEyes here or Data360....
- Perceptually or cognitively too hard to understand: Much of visualization is designed not for the masses, but rather for the specialist. A challenge is to design visualizations that can be easily consumed or laypeople can be easily trained to use. Think about treemap being used on SmartMoney.com. How many have really learned to look at it? That's a crucial question to study.
- Design Research: I call this one the "Beyond-Tufte problem". How do we get beyond having geeks and gurus tell us how to design visualizations? We need a rigorous HCI design research approach to visualization that encompasses ethnographic studies, user-need analysis, iterative design, and real evaluations beyond cute lab confirmation studies.