and details-on-demand capabilities for parallel coordinates.
The literature covering parallel coordinates is vast and
covers multiple domains as recently surveyed by Heinrich
and Weiskopf [117].
EDEN extends the classical parallel coordinates axis by
providing cues that guide and refine the analyst’s exploration
of the information space. This approach is akin to
the concept of the scented widget described by Willett
et al. [118]. Scented widgets are graphical user interface
components that are augmented with an embedded
visualization to enable efficient navigation in the information
space of the data items. The concept arises from
the information foraging theory described by Pirolli and
Card [119], which relates human information gathering
to the food foraging activities of animals. In this model,
the concept of information scent is identified as the ‘user
perception of the value, cost, or access path of information
sources obtained by proximal cues’ [119]. In EDEN,
the scented axis widgets are augmented with information
from automated data mining processes (e.g., statistical
filters, automatic axis arrangements, regression mining,
correlation mining, and subset selection capabilities) that
highlight potentially relevant associations and reduce
knowledge discovery timelines.
The parallel coordinates plot is ideal for exploratory
analysis of materials science data because it accommodates
the simultaneous display of a large number of variables
in a two-dimensional representation. In EDEN, the
parallel coordinates plot is extended with a number of
capabilities that facilitate exploratory data analysis and
guide the scientist to the most significant relationships in
the data. A full description of these extensions is beyond