Integrative Visual Abstraction of Molecular Data

In the visual arts, abstraction is a way to emphasize the structure, the atmosphere, overall shapes, or the dynamics of a depicted scene. In illustration, abstraction further assists us to convey important concepts, methods, or various physical phenomena. Computer visualization can thus use abstraction to convey insights that would otherwise be buried in complex data sets. But to this date we only have a rather implicit understanding of what this abstraction really is. By our intuition, abstraction is a process of removing the concrete, those specifics that relate to one particular object or subject---but it is hard to characterize abstraction more specifically. In our research project we thus aim to characterize abstraction principles so that we can use them more effectively in data visualization. Such abstracted visualization becomes especially useful in cases where various types of information are to be conveyed simultaneously. Without a careful visual resources management the resulting visual encoding would result into a non-understandable clutter: a multitude of visual information that is fighting for a viewer`s attention. Such visual clutter comes in many scenarios, even in daily tasks like navigating with a GPS device showing all possible points of interest at the same time in a dense urban area. In our project we want to challenge the gained knowledge about abstraction forms on complex molecular data that encode multiple aspects at multiple spatial and temporal scales, while still focusing the depiction on the essential elements to help researchers to analyze the data.

In many aspects of our modern lives we face the problem of having to make sense of large amounts of data. This applies to scientists trying to make sense of their experiments and simulations, to bankers and traders trying to understand the dynamics of the financial markets, and to large enterprises which need to understand the principles behind demand and supply-to name just a few examples. The basic problem in all these cases is that people need to identify important and/or unexpected features in large simulated or captured datasets. Visualization is the domain that facilitates this process of sense-making by dramatically simplifying the process of obtaining an understanding of the data-by representing data visually and thus amplifying people's cognition. This inherent capability of (good) visualization techniques to amplify human cognition, however, is no longer enough to be able to make sense of today's huge datasets. To be able to see the essential aspects we need dedicated mechanisms that abstract away the (unnecessary) detail to, in turn, allow the user of the visualization to focus on the important elements. The crucial problem in this context is that it is impossible to know what is important and what is not in a general way-importance changes based on the research question, on the application domain, on the data size, on the user, on the specific situation, etc. Visualization technology therefore needs to support dynamic change of visual abstraction of the data to reflect these contextual changes. The fundamental research challenge in visualization for us is to get an understanding of what (visual) abstraction really is, what it means, how it can be controlled, and how it is, can be, and should be used in visualization.