Project: Navigating Abstract Data Spaces With Fish-eye Lenses

An information space is a broad term used to describe everything from a web site structure to a network schematic to a single text document. Navigating a small information space is easy. You can see everything at once, and it is clear how what you're looking at fits into the overall structure. But as the size of our data structures continues to grow, the space on our screens does not grow to match it. The "window" through which we look at the structure is limited both by technology of the screen and by the human visual processing capacity.

Consider how much the average user's display has changed in the past twenty years; perhaps from 640x480 on a 12" screen to1024x768 on a 17" screen... less than doubled. Consider how much the human capability to assimilate information has changed... not much. Now consider how much not only average RAM and hard drive space have increased, but the amount of information accessible from the Web. There is a considerable difference!

So what part of the structure should we show? Obviously, we want detail, but we also want a sense of the overall structure and our location within it. These two levels of information are called "focus" information and "context" information. If we provide a zoom window onto the structure (consider the window of a typical word processor, for example) we get maximum detail for the selected area, but lose the sense of where that area is in the larger structure (focus but no context). If we present the whole structure at once, it is either summarized or compacted to fit on the screen, and we can no longer see any detail (context but no focus).

Proposed Research

A fisheye lens attempts to provide both focus and context information at once. In the world of photography, a very wide angle (fisheye) lens used at a short distance shows things near the center of the lens in high magnification and detail. At the same time, however, it shows the surroundings in decreasing magnification and less detail as they get further away from the center.

Why haven't fisheye lenses swept the visualization world? Distorted views can initially be confusing. Consider the illustrations above. The figure on the left is an undistorted graph; the right shows a distorted view. Note how the spatial relationships of all the nodes have changed, and try to find the focus node in the left graph. While all of the graph is always shown, its appearance is very different. Does this matter?

In most navigation tasks, whether they are in the real world or in an abstract space, a person typically first learns landmarks. In a graph like those above, there are several kinds of things that could be considered landmarks; the positions of the nodes on the screen ("The one in the upper right corner"), the appearance of the nodes ("The only purple one") or the edges between the nodes ("The nodes that form a long single file line").

What this research is trying to determine is what kind of things get used as landmarks with different degrees of distortion of the graph. We hope the results will provide guidelines for designing fisheye representations that can be easily navigated.

Contact Amy Skopik if you would like more information on this project.

Participants

Amy Skopik Carl Gutwin
University of Saskatchewan

Publications

Improving Revisitation in Fisheye Views with Visit Wear
Skopik, A., Gutwin, C. (2005), Proceedings of the ACM Conference on Human Factors in Computing Systems, 771-780.
Improving Memorability in Fisheye Views
Skopik, A. (2004) M.Sc. Thesis, Department of Computer Science, University of Saskatchewan.
Fisheye Views are Good for Large Steering Tasks
Gutwin, C., Skopik, A. (2003), Proceedings of the ACM Conference on Human Factors in Computing Systems, 201-208.
Finding Things In Fisheyes: Memorability in Distorted Spaces
Skopik, A., Gutwin, C. (2003), Proceedings of Graphics Interface, 47-56.