Designers Jörgen Abrahamsson and Johan Wastring met up with Jonas Löwgren at the Medea studio on December 14, 2010, to continue our little ongoing chat on ideas and possibilities in information visualization and interactive visualization.
Jörgen is currently working on new ways to communicate city welfare statistics, using static visualizations to enhance the conventional printed reports from the City council. (Click the image for a larger version, as always.)
The semi-circular diagrams, for instance, are inspired by the rings of a cut log and illustrate general welfare level over the course of the entire lifespan of many people at once. Each line from the center to the periphery represents one person.
We spent some time on Jörgen’s ideas of using simplex graphs or triagrams to visualize tripartite proportional distributions. Basically, the idea is that each data point is plotted inside an equilateral triangle. Each side of the triangle corresponds one of the three interdependent variables, and the perpendicular distance from the point to each side corresponds to value of the data point in that variable.
A related idea that came up was to use a tetrahedron (a pyramid, basically) in a 3D space to diagram four-part variables.
Johan has invested a lot of time recently in playing around with graph data structures and ways of visualizing them in 3D space. Learning to visualize graph data is a bit of a technical feat; Johan’s journey has gone from Prefuse and Flare, which are competent drawing libraries but do not offer much in terms of graph computation algorithms, to Jung and JGraphT which offer much more in terms of algorithmic muscle.
The demo that Johan shows here is an exploration of how interaction and 3D can help the user gain insights from large graph structures, something that seems to be a notoriously difficult task in information visualization.
Our discussion touched on the relation between mathematical methods and human judgment. Graph data structures lend themselves to quite a bit of computation, and there might even be a temptation to push the automation too far in an effort to wring meaning from the data. The challenge is, of course, to find the sweet spot in dividing the labor between computing graph properties and presenting data in ways that empower the user to explore and gain insights.
In a recent article on infosthetics entitled The End of Network Graph Hairballs, a compelling point is made that the typical densely populated network graphs are difficult to understand and compare, and that the layout rules are difficult to predict even to the point where it is hard to tell whether what you think you see is an meaningful relation in the data or merely an artifact of the layout algorithm. We tended to agree with this observation, even though we felt that the solution proposed in the article (a technique called linear network layout) would hardly solve the problem for good.
To wrap up the meeting, Jonas gave a status report of the ongoing work in the national strategic foresight panel which is commissioned by research funding bodies such as KK-stiftelsen, Vinnova and Vetenskapsrådet to suggest promising directions for future research directions in visualization.
Briefly, the panel uses the concept of “perceptualization” to emphasize the significance of modalities other than the visual. The work is at a stage where three main directions have been identified: basic research into perception and perceptualization, support for qualified collaborative problem-solving, and everyday perceptualization. The goal is to have a result in February 2011 where a number of more specific directions for research, development and innovation are identified.