ANGULAR TREEMAPS - Information Visualisation: Techniques, Usability and Evaluation - page 84

The value of data includes its attributes and elements. We can view a
value distribution or density in 2D, 3D or higher-dimensional visual
spaces. For data sets with lower dimensions, a typical example of a
visualization format is a scatter plot [2]. To visualize multiple data values,
we can build graphical attributes into scatter plots (e.g. including position,
size, etc.). For higher dimensional spaces, parallel coordinates are typically
used.
In practice, many information sources are organized in hierarchical
forms. For example, the organizational structure of a file system, structure
of a classification system, structure of an organization, taxonomy of
objects, and many more are hierarchical in nature. These hierarchical
structures are often complex, with thousands or millions of elements and
relationships. As a result, it is crucial to provide effective interactive
visualization methods for hierarchical data, with the capability of
exploring different levels of granularity, while emphasizing the importance
of the data for analysts as part of the knowledge discovery process.
Space-filling visualization has been a successful method for visualizing
large hierarchical data sets with attributed properties. This approach uses
an enclosure approach to represent tree structures, ensuring that all nodes
and their sub-hierarchies are located inside their “father's” display region.
It can provide a visual presentation of global patterns of the overall data
structure in a compact display. This technique maximizes space efficiency
by partitioning the display area into nested sub-areas and assigning them
to geometrical regions that represent subsets of the dataset in the enclosed
display area. Space-filling techniques, especially Treemaps have been
commercialized successfully into many domains, such as finance analysis
[3], sports reporting [4], image browsing [5], and software analysis [6].
However, most of the existing Treemap algorithms are based on axis-
aligned rectangles. Gestalt research and Geon Theory [7] have shown that
humans have a tendency to seek out an object's edges, which they can
quickly detect when one shape is different from another. Hence, restricting
Treemap visualizations to vertical-and-horizontal rectangular blocks,
constrains the human capability of object recognition, due to the same
fixed angles of shapes. This assertion is supported by theory rooted in
empirical perception data. User studies [8] have investigated the effect of
shape variation of data elements in a series of control experiments that
compared combined treemaps with traditional rectangular treemaps. This
empirical outcome demonstrated the effectiveness of combined treemaps,
reinforcing the idea that shape supports human cognitive thinking in
certain scenarios of visual data analysis [8]. Therefore, variations in angles
and/or shapes of nodes positively direct graphic perception and insight