Content uploaded by Oh-Hyun Kwon
Author content
All content in this area was uploaded by Oh-Hyun Kwon on Oct 21, 2016
Content may be subject to copyright.
Strata Treemaps
Junghong Choi*, Oh-hyun Kwon† and Kyungwon Lee‡
Division of Digital Media, Ajou University
*sakool@ajou.ac.kr, †kwonoh6@ajou.ac.kr, ‡kwlee@ajou.ac.kr
Figure 1. Strata Treemap: visualizing directory structure
1. Introduction
Treemap is a method of visualization to recognize a hierarchical
structure of data using the size and arrangement of nested rectan-
gles. A limitation of a treemap is the difficulty to discern the struc-
ture of a hierarchy. Several approaches have been proposed to
improve the visibility of the hierarchical structure. These ap-
proaches involve the use of a border or padding to emphasize the
hierarchical structure. However, this leads to a disparity between
the node weight and the relative node size.
Strata treemap, a new approach to treemap presentation that im-
proves the visibility of the hierarchical structure without distorting
the node sizes. Strata treemaps are based on Voronoi treemap.
Strata treemap use the surface of a sphere instead of the traditional
two-dimensional plane. For each node in the hierarchy, we extrude
it along the its surface normal and stack up on the its parent. The
result is a three-dimensional shape that consists of stacked blocks
on a sphere. Each block represents a node in a hierarchy and its
top surface area represents the quantitative value of the node.
Additionally, original treemaps use the metaphor of hierarchy as
containment. Accordingly, it is difficult to observe the hierarchy as
a series of higher or lower levels. However, the stacked structure
of a Strata treemap makes this easy.
2. Approach
Our goal is simple: improve the visibility of the hierarchical struc-
ture without distorting the node sizes or eliminating them. We
created spaces between the nodes (siblings) to facilitate easy rec-
ognition of the hierarchical structure.
Figure 2 shows our idea in a simple manner. The farther a cer-
tain spot on the surface of sphere is from the center, that is, the
longer the radius, the larger the sphere. We utilized this feature of
a sphere but did it the other way around. Initially, we formulated
treemaps on the surface of a sphere. We then extruded the surface
of the nodes along the its surface normal, not by extending the
radius of the sphere (Figure 2-1). In this way, empty spaces be-
tween the nodes are naturally generated while the size of a node
does not change (Figure 2-3). Moreover, labels can be attached
onto the lateral sides of the nodes without creating additional
spaces for labeling.
Figure 2. Basic Idea of Strata Treemaps
3. Algorithm
We made an algorithm based on weighted centroidal voronoi tes-
sellation. The basic structure of the algorithm is similar to the
recursive structure of the original treemap layout algorithm. We
divided the surface of the node to get the surfaces of its children
nodes. For each children nodes we extruded it along the normal
direction of its surface and stack up on its parent node.
4. Results and Conclusions
As a final example, Figure 1 shows a Strata treemap illustrating a
directory structure of openFrameworks (open source c++ library).
This hierarchy contains 9,892 nodes, including 7,512 leaf nodes,
and 14 depths. Top surface area of each nodes refer to revision
count of files. Color used as classification of root directory. Labels
are attached onto the lateral sides of nodes without creating addi-
tional spaces for labeling. Most of the structure of this hierarchy is
easily visible. Moreover, as expected, there is no distortion or
missing nodes.
The Strata treemap is a new type of data visualization scheme
which maintains the size and emphasizes a hierarchical structure.
In existing methods, node sizes are subject to distortion for better
identification of the hierarchical structure. In Strata treemaps,
however, distortion of the node sizes is no longer necessary.
Therefore, the nodes in a Strata treemap can be compared with
each other regardless of the portion they are in. Furthermore, hier-
archical structure is easily recognized, as each node is stacked on
its parent node.
Our future work will improve user interface with the strata tree-
maps that ‘strata treemap browser’. It will allow the user to peel
back hierarchical layers, transform nodes, filter and navigate. It
would be similar concept to ‘google body browser’ that peels back
anatomical layers, zooms in, and navigate to parts.
5. References
HAO, L. AND JAMES, F. 2008. Cascaded treemaps: examining the
visibility and stability of structure in treemaps, In Proc. of Graphics
Interface 2008, Canadian Information Processing Society, 259-266.
MICHAEL, B. AND OLIVER, D. 2005. Voronoi treemaps, In Proc. of
the IEEE Symposium on Information Visualization 2005, IEEE
CS Press, 49-56
ZIEMKIEWICZ, C. AND KOSARA, R. 2008. The shaping of informa-
tion by visual metaphors, IEEE Transactions on Visualization and
Computer Graphics, IEEE Educational Activities Department,
14(6):1269-1276
