VISUALIZING THE EVOLUTION OF SOFTWARE SYSTEMS USING THE FOREST METAPHOR - Information Visualisation: Techniques, Usability and Evaluation - page 129

Ghandar et al. [23] propose a jigsaw puzzle metaphor, where each
component of a system is modelled as a piece of a jigsaw puzzle. The most
significant characteristic of this metaphor concerns the visualization of
system complexity that is represented as a pattern on a puzzle piece's
surface. The metaphor does not provide a view at the class granularity
level, namely it does not provide any fine-grained representation to
support software comprehension. This is one of the most important
differences with respect to our proposal.
Several differences exist between our proposal and the approaches
discussed above. The most notable is that our proposal offers a proper
representation for methods, attributes, and source code comments. A
possible drawback that affects our metaphor is that it could appear too
complex for maintainers who are not properly trained. This issue is
directly connected to the range of types of information our metaphor is
able to visually summarize. Another difference is that all the approaches
above scarcely, at best, provide support for the understanding of software
evolution.
In [24] a 3D visual representation for analysing a software system's
release history is proposed. The approach is based on a retrospective
analysis technique to evaluate architectural stability, based on the use of
colours to depict changes in different releases. Tu and Godfrey [25]
propose an approach that makes an integrated use of software metrics,
visualization, and origin analysis. An approach is suggested in [26] to
analyse how changes appear in software based on the notion of history.
The authors also introduce a tool for the visualization of histories in
evolving class hierarchies. The main difference between these approaches
and ours is that we consider all the releases of all the source code artefacts,
while these approaches consider snapshots of the system. Furthermore, our
proposal is based on a natural environment, and is able to show large- and
small-scale understanding of evolving software.

The Metaphor

We propose an approach that visualizes object oriented software systems
as a 3D rendering of forests of trees. The rationale for adopting the forest
metaphor and defining our approach is related to large- and small-scale
understanding concerns and can be summarized as follows:

x a tree visualization summarizes complex information in a natural

way;