Seeing Science

1 Seeing Science Alyssa A. GOODMAN1,2 The ability to represent scientific data and concepts visually is becoming increasingly important due to the unprecedented exponential growth of computational power during the present digital age. The data sets and simulations scientists in all fields can now create are literally thousands of times as large as those created just 20 years ago. Historically successful methods for data visualization can, and should, be applied to today’s huge data sets, but new approaches, also enabled by technology, are needed as well. Increasingly, “modular craftsmanship” will be applied, as relevant functionality from the graphically and technically best tools for a job are combined as-needed, without low-level programming. 1. Introduction The essential function of data visualization is to offer humans a way to see patterns in quantitative information that would otherwise be harder to find. Many people today believe that computers can always find these patterns as easily, or more easily, than people can. The people who do not believe computers have this power fall into two groups: researchers who strive to create tools as good as humans, and small children (who have not yet been indoctrinated to believe that computers are superior computers to humans in all ways!). The most productive research in data visualization today is focused on developing technology to augment the human ability to find patterns. 2. History Before the introduction of the computer into science, data visualization took two forms: 1) hand- drawn sketches made by researchers themselves; and 2) professionally-drafted illustrations. Some “conventions” for making these drawings did develop (e.g. Cartesian coordinates), but the makers of early scientific drawings were free to draw upon or create whatever tools and rubrics were most appropriate to their tasks, conventional or not. As computers entered the picture, several important changes took place. First, on the upside, the amount of data scientists could generate and analyze began to rise very rapidly, and the alternatives available for how to display it (e.g. animation, 3D graphics) began to expand. On the downside, the tools that were developed to put data visualization into the hands of the scientists themselves typically offered nowhere near the level of flexibility and craftsmanship that the combination of hand-drawing and professional draftspeople could. As a simple example, think about how easy it is for a person to write a name along a curving river or street in a map (Figure 1), but how much harder it is to get a computer to do that just as well. Today, the very best tools available for data analysis and visualization are being developed with attention to the kinds of graphical details and functionality that the work of draftspeople used to add to science. Below, I argue that what is needed now is for high-craft tools to be made modular and interoperable enough so that scientists can combine the functionality offered by various systems into ones where “modular craftsmanship” is possible. 3. Data • Dimensions • Display Formally, we can frame visualization challenges by thinking about interactions amongst data, dimensions, and display. Some data to be visualized arise from continuous functions (e.g. fitting), others come from discrete measurements (e.g. observational/experimental data). Some data sets are inherently large and others small. Most data sets have either an inherent dimensionality, or dimensionality imposed when a choice is made about what quantity/quantities are to be explored/displayed as functions of others. For example, brain imaging data is often three-dimensional, but is often displayed as a series of two-dimensional slices. Oftentimes, it is the nature of a display mode (e.g. monochrome vs. color, paper vs. electronic, static vs. dynamic, etc.) that sets boundaries on what data are displayed with what dimensionality. The word “dimensionality” should not be taken too literally. In some cases, such as medical, geospatial or astronomical data data, there are natural To appear in: Proceedings of the International Festival of Scientific Visualization, held in Tokyo, Japan, March 2009. Publisher will be Universal Academy Press.　　　 1 Professor of Astronomy & Founding Director of the Initiative in Innovative Computing, Harvard University 2 Scholar-in-Residence, WGBH Boston Fig.1. An historical map of Edo (1844-48). Notice the craftsperson’s attention to orientation in the labeling, and the beautiful details of illustration. [1] 2 coordinates, with a natural number of dimensions (typically 3 plus time) in which one can display sensed quantities. Often, that kind of “natural” display is particularly useful. But, even in fields offering what seem “natural” combinations of dimensions, there are often abstract combinations of “dimensions” (e.g. a 3D plot of

…(Full text truncated)…

This content is AI-processed based on ArXiv data.