[Back]

I. Viola:
"Declarative Visualization";
Talk: SCCG 2013 - 29th Proceedings Spring Conference on Computer Graphics, Smolenice, Slovakia (invited); 2013-05-01 - 2013-05-03.

Visualization algorithms are nowadays formalized in an imperative manner, i.e. the algorithm is explicitly executed on input data and dictates a determined visualization outcome. The efficiency of such an algorithm is measured by means of the computational performance, data-scalability and user studies. In my talk I will speculate on a novel theoretical concept for the development of new visualization methodology that becomes ultimately declarative and algorithm-free, by moving the user study from a validation stage into the center of the iterative design stage. Initial visualization from input data is considered as the first design draft, which will undergo several revisions. This draft can be achieved by executing a traditional imperative algorithm or it can even be hand-crafted by a skilled illustrator. A consequent user study of initial visualization will trigger computational synthesis of a new, quantitatively more effective visualization technique. The visualization designs developed through several iterations of the study-redesign cycle will become declarative, aiming at optimally satisfying the purpose of the visualization, instead of explicit execution of algorithms on the input data. The declarative component will be specified by collected user statistics from completing certain perceptual or cognitive tasks. The user statistics will be analyzed for systematic trends in human perceptual and cognitive performance. These trends will form a basis for visualization redesign. Final satisfactory visualization will evolve over several design iterations.