| TREE | |
. | |
| Workshop 2 | |
||
|
Ordered
Treemap Layouts
|
|
|
Ben Shneiderman
ben@cs.umd.edu Univ. of Maryland Martin Wattenberg wattenberg@smartmoney.com Smartmoney & Columbia University |
|
Treemaps, a space-filling method of visualizing large hierarchical data sets, are receiving increasing attention. Several algorithms have been proposed to create more useful displays by controlling the aspect ratios of the rectangles that make up a treemap. While these algorithms do improve visibility of small items in a single layout, they introduce instability over time in the display of dynamically changing data, and fail to preserve an ordering of the underlying data. This paper introduces the ordered treemap, which addresses these two shortcomings. The ordered treemap algorithm ensures that items near each other in the given order will be near each other in the treemap layout. Using experimental evidence from Monte Carlo trials, we show that compared to other layout algorithms ordered treemaps are more stable while maintaining relatively low aspect ratios of the constituent rectangles. A second test set uses stock market data. |
|
Trees and
cushions slides
|
|
|
Jack van
Wijk
vanwijk@win.tue.nl Dept. of Mathematics and Computer Science Eindhoven University of Technology |
|
| The Treemap is a very elegant method to display tree structures in a very compact way. Since the late nineties we have tried to improve it further. The structure of large, balanced trees is sometimes hard to detect. The addition of hierarchical cushions gives an improved perception of the structure. We have used this method to visualize the contents of hard disks, a freeware tool can be downloaded from www.win.tue.nl/sequoiaview. Another extension concerns the elimination of elongated rectangles. A new algorithm has been developed to size and position subrectangles within a rectangle. An overview of these methods will be presented. Recently we have worked on other applications of cushions (Voronoi diagrams, contour plots) and also on other tree visualization methods. The use of methods from botanical modeling leads to interesting results. Some first results will be shown. Finally, topics for future research concerning treemaps are presented. |
|
Crossing
the Chasm: Smartmoney's success story with the marketmap
|
|
| SmartMoney's Map of the Market has become the treemap standard for viewing of stock market data on the Internet. This presentation will provide an in-depth look at the research and development effort that went into launching the Map of the Market on the Smartmoney.com Web site and an examination of how SmartMoney's users interact with the Map. Mr. Frons will also demonstrate some of SmartMoney's latest innovations in treemap visualization. |
|
Quantum
Treemaps for a Zoomable Image Browser
|
|
This paper describes PhotoMesa, an application that supports browsing of large numbers of images. PhotoMesa uses a Zoomable User Interface with a simple interaction designed for novices and family-oriented use. Images are grouped by directory or other metadata, and are layed out using a Quantum Treemap algorithm, a variation on existing treemap algorithms designed for laying out images or other objects of indivisible (quantum) size. Quantum Treemaps build on the Ordered Treemap algorithm, but guarantees that every generated rectangle will have a width and height that are an integral multiple of an input object size. |
Treemaps have so far been used to display static views of hierarchical information mainly for exploratory purposes. Most of the time, the data at hand are multivariate and the user is often interested in seeing alternative views of the same hierarchy by changing the visible variables used for weighing or filling. Consequently, some areas of the display shrink while other areas expand so much so that the transition between the alternative views of the same hierarchy are impossible to follow. This paper describes early attempts at using animation with two treemap layouts. |
|
Modifiable
Treemap Containing Boxes of Variable Size
slides
|
|
| Our algorithm extends the initial Treemap to improve the final shape of elements that are often very long rectangles. The new algorithm called Fieldmap optimizes the labels displayed on each element. As our algorithm tends to confuse the user on the exact location of one element in the tree, we propose new features to highlight the selected elements (graphic and sonic cues will be discussed). We implemented a java version of our algorithm (applet: http://iihm.imag.fr/vernier/indexParent.html) to manipulate the traditional hierarchy of files. The drag drop moves, in one action, the selected files between two far locations of the tree. A combination of hue/brightness is used to transcript file age and file type. Two interactive legends are used in coordination with the Fieldmap to colorize the data. Each legend is used by direct manipulation to highlight the FieldMap. At last we propose to replace, in a sub part of the Fieldmap, the layout by a indented list on user demand. |
Visualizing Social Cyberspaces |
|
|
Marc A.
Smith,
masmith@microsoft.com Research Sociologist Microsoft Collaboration and Multimedia Group http://netscan.research.microsoft.com Andrew Fiore, atf2@cornell.edu Cornell University |
|
| Interfaces to social cyberspaces, such as discussion boards, email lists, and chat rooms, present limited if any information about the social context of the interactions they host. Basic social cues about the size and nature of groups are missing, making discovery, navigation, and self-regulation an increasing challenge as the size and scope of these spaces expand. However, it is possible to generate such information by data mining log files of message bases created through such systems. We present here results of the Netscan project to data mine and visualize Usenet, one of the largest collections of social cyberspaces. Our initial goal is to provide an overview of the range of variation in Usenet and to highlight indicators and frequencies of different types of groups. This information can drive enhanced user interfaces that reveal patterns and dynamics of interaction not visible in standard interfaces. In particular, treemap, suggest a useful direction given their native ability to visualize hierarchical structures and map on to them multiple layers of additional data. We have had some success in producing large scale visualizations of Usenet (see (386K): http://www.research.microsoft.com/~masmith/all_map.jpg) that illustrate the proportional size and growth of all newsgroups. The challenges for future work include making successive tree maps more easily comparable, allowing time series data to be intuitively displayed; finding new methods for making proximity more meaningful (for example basing it on crossposting relationships), and performance issues in providing highly complex interactive maps through web interfaces and rendering technologies. |
|
A Free
Java Library for Treemap Visualization slides
|
|
|
Christophe
Bouthier.
Computer Science PhD Bouthier@loria.fr ECOO Team (http://www.loria.fr/equipes/ecoo) Nancy - France |
|
| To be widely used, a widget should not compel developers to reimplement it every time. That's why widget libraries exist. They provide developers with easily configurable and ready-to-use widgets. In order to promote treemap usage in applications, we have developed a treemap java library : give it tree-structured data set and you will get configurable treemap views, updated when data changes. You could choose the way to compute the drawing or the size of nodes, if you want a classic or squarified treemap, and the size of the nested border if any. The library is implemented in Java, using Swing and design patterns, and is multi-threaded. Finally, it is an open-source project, put on sourceforge (treemap.sourceforge.net), free to be used, distributed and modified. |
|
Musings
on Treemaps slides
|
|
|
Our group has
quite a bit of experience working with Treemaps, dating back to a UNIX
implementation of the Treemap algorithm for browsing file structures.
We built this shortly after the original Treemap papers were published
in order to understand the technique better and to have a useful tool
on our workstations. The tool is still in use today and it has proven
quite helpful over the years. Recently, we developed a space-filling,
radial, Treemap-like system called SunBurst and we compared it to a Treemap
implementation in an empirical study of people performing file browsing
tasks. I will briefly present the results of that study and highlight
the seeming affordances of each tool. Finally, we've developed some animated,
zooming techniques for focusing on extremely small items in a SunBurst
display, and I will speculate about how similar techniques could be applied
to Treemaps to make them even more powerful.
|
|
Adding
Dynamic Queries to Interactive Treemap slides
|
|
|
K Babaria,
B Bederson, S Betten, J Blowitski, C Plaisant, B Shneiderman
HCIL, Computer Science Department University of Maryland |
|
| Treemaps are space-filling visualization for hierarchical structures that are extremely effective in showing attributes of leaf nodes by size and color-coding. Treemaps enable users to compare sizes of nodes and of sub-trees, and are especially strong in spotting patterns. Dynamic Queries technique is a visual alternative to SQL for querying a database. It involves an interactive control of various parameters using sliders, buttons etc and it generates rapid, animated visual display of database search results. This summarizes work done at HCIL of adding Dynamic Queries to treemaps. |
|
Treemaps
for Design Knowledge Visualization and Reuse slides
|
|
|
Peter
Demian,
Ph.D. Candidate Prof. Renate Fruchter, Director of Project Based Learning Lab, Department of Civil and Environmental Engineering Stanford University |
|
|
The proposed architecture for
CoMem is based on the principle of "overview first, zoom and filter, and
then details-on-demand." We propose to use the treepmap technique to visualize
an overview of the corporate memory, and filter items using dynamic querying.
The filtering can be based on relational attributes, on a similarity measure
generated using latent semantic indexing, or on the idea of "semantic
zooming". |
|