A Spreadsheet Interface for Visualization Exploration
Without effective interfaces to represent the process and results of visualization, a user cannot properly utilize the underlying techniques to extract information from their data. A spreadsheet-like interface to visualization can be used to store and present the data exploration process itself. This information can be shared and reused, enabling users to build upon past work without repeating themselves. The interface does so via the following properties:
- A tabular structure that encapsulates the visualization.
- Operators upon parameters and results to analyze and manipulate the visualization.
- An interpreter that can control the visualization process at a lower level.
The first figure demonstrates the first two aspects. The spreadsheet presents a tabular view of its underlying data. It is a movable, scalable window into visualization space. By manipulating the visualization parameters, the user changes the position and size of this window. The spreadsheet displays a planar projection spanned by two axes of the space. In the example, the rows represent color maps and the columns opacity maps for volume rendering. Unlike conventional interfaces, the correspondence between parameter value and result is clearly indicated by the rendering of the parameters in the interface. For the non-displayed parameters, the user selected default values to be used. The rendered value for a cell then combines the displayed row and column parameter value with the defaults for the non-displayed parameters to produce the image. By using the spreadsheet-like interface, the visualization exploration becomes a process of maneuvering the spreadsheet window through the parameter space.
Operators upon parameters and results are another important element of the interface. In the previous example, the user is examining the positive and negative vorticities of a turbulent flow in the first and second row/column respectively. To view the flows simultaneously, a set union operator can be applied to the parameter values, generating the final combined image in the bottom right cell. The operators allow more efficient exploration of the data by allowing the user to build upon previous results.
The next figure demonstrates a spreadsheet-driven visualization; in this case, the goal was to display separate skin and bone surfaces for a foot data set. First, the user added two opacity maps which highlight the desired surfaces. The tabular organization of the spreadsheet allows the two images to be easily compared side-by-side. After changing the row parameter to display view positions, the user selected a view to display the front of the foot. This position was then selected as the new default. Afterwards, the user returned to modifying color maps; only images utilizing the new position were displayed. Two new color maps were added, the first a false-color map highlighting the differences in value on the surfaces and the second a color map to display a flesh-like tone for the skin and white for the bone; the latter color map was selected as the new default. Finally, the final images were generated by displaying and adding a new zoom factor value. If the user wanted, they could change the default color map or view position to examine alternate zoomed images.
The spreadsheet interpreter allows manipulation of the exploration process at a lower level. The interpreter can access any parameter or cell value or operator and is not limited to the values currently displayed. Expert users can also create templates to assist others in common exploration tasks. The history of the exploration process can also be displayed via animation and shared with others via an XML representation.
Our interface assists in visualization process in two ways. First, the structure of the spreadsheet provides an organized means of exploring and manipulating the space of visualization parameters. Secondly, the interface eases collaboration as it communicates all the steps used to the find desired results. This history can be useful in exploring alternate paths at a later time.A paper describing this has been published in the IEEE Visualization 2000 Conference Proceedings. The reference is:
T. J. Jankun-Kelly and Kwan-Liu Ma, "A Spreadsheet Interface for Visualization Exploration," in Proceedings of IEEE Visualization 2000, Salt Lake City, UT, October 8-13 2000, IEEE Computer Society Press, Los Alamitos, California. [Paper (PDF)]
An expanded version has been printed in the July--Septemeber 2001 issue of IEEE Transactions on Visualization and Computer Graphics:
T. J. Jankun-Kelly and Kwan-Liu Ma, "Visualization Exploration and Encapsulation via a Spreadsheet-Like Interface," IEEE Transactions on Visualization and Computer Graphics, 7(3), IEEE Computer Society Press, 2001, pp. 275--287. [Paper (PDF)]
Also available: Overview (PDF)
Contact: T. J. Jankun-Kelly, Kwan-Liu Ma
