This research was developed during my work experience in Rome, Italy.
It was an effort towards clearly establishing a link between visual techniques and
time-dependent data.
The temporal dimension is
becoming more and more important in modern database applications, such as
medical and biographical records, multimedia systems, geographic information
systems, scientific and statistical applications, etc. Such applications are
spread over the world and need to be accessed by several people with different
expertise and needs. This is driving the development of ad-hoc visualizations
and interaction mechanisms which address the specific requirements of
time-dependent information. It is now recognized that the initial approaches,
just considering the time as an ordinal dimension in a 2D or 3D visualizations,
are inadequate to capture the many characteristics of time-dependent
information.
However, visualizing time-related data is more than
ordering them along an axis. Indeed, extensions of existing visualization
techniques or new techniques have mainly focused on time-oriented interactive
exploration and visualization of large datasets. Note that most of them address
a particular notion of time (e.g. linear or branching time, discrete or
continuous time) in domain-specific applications, since it is difficult to
represent the different aspects of temporal data in a single visual structure, we have presented a survey of
the main visual techniques for interactive exploration of time-oriented
information. Such survey is based on a object-oriented
framework for temporal data models as a unifying realm for various features
found in several temporal data models. Furthermore, we characterize
time-oriented visualizations as supporting snapshot or slice
views, which correspond to the visualization of an instantaneous fact (valid at
a single time instant) or an historical fact respectively.
The combination of the snapshot/slice view with the
classification provided by the above framework gives rise to the following categorization
of the visual techniques:
Slice Visualization
Which corresponds to a
visualization of valid history, i.e., a visualization of one or more entities
(and their attributes and relationships) valid at discrete (continuous)
instants or intervals in a linear
order. Most of the systems which use this technique propose to visualize
temporal data through interactive 2D timelines. More recent approaches aim at
better facilitating the user-interaction through advanced visual techniques,
such as visual distortion, 3D visualization/animation, visual metaphors and
automatic generation/display of timelines. Furthermore, there are systems that
are tailored to specific applications, such as geographical information
systems, mobile computing, VR-systems, allowing the visual exploration of video
and spatio-temporal data.
Examples of Interactive/Innovative Timelines
|
|
|
|
|
|
Periodic-Slice Visualization
Which corresponds to a visualization of valid history
at specific discrete (continuous) patterns of time (calendar). Visual
techniques have been proposed explicitly dealing with the visualization of
periodic patterns in historical data, mainly used in scheduling applications,
such as the electronic calendars
which have arisen as enhancements of paper calendars, since they exploit
several visual clues, such as highlighted view of related events, graphical
symbols as patterns for cyclic events, etc. Other visual techniques explore
periodic patterns found in data mining, where time is considered a dimension in
a multidimensional dataset (time-series
data) and the change of data values is tightly associated with a specific
pattern of time. The spiral visualization and other visual techniques have been
adopted for visualizing time-series data.
Examples of Visual Calendars/Spiral Visualization
|
|
|
Snapshot Visualization
Most of the visual techniques we have recalled in the
above provide the user with the interactive exploration of information across
several time periods in a linear or branching order, to identify temporal
trends/evolution of one or more objects. However, in some applications, such as
personal histories, image sequences, etc, an additional functionality is needed
that allows the snapshot visualization of an information, i.e., visualization
of data valid at a single discrete (continuous) instant or interval. The
snapshot visualization is very useful in presenting current and past
instantaneous facts, highlighting a facet of a current view or a past state of
an information.
Examples of Snapshot Visualization of an Information
|
|
According
to such a categorization, we discussed the characteristics of the various
visualizations and presented examples of systems implementing them. Moreover,
he have summarized the main visual (periodic patterns, snapshot view, multiple
calendars, visualization of complex entity and entity relations, user-defined
display, patterns/trends (not periodic), focus+context, etc) and interaction
(time-traveling mechanism, overview, zooming, filter, details-on-demand)
features of the systems and issues related with visual query modalities.