Database Reference
In-Depth Information
potentially reduce the amount of ad hocness in the field. The key is to define meas-
ures of information transfer, content, or loss at all stages of the pipeline as a means of
assessing our progress in the development of new visualization techniques and en-
hancement of existing ones.
4
Formal Models for a Science of Information Visualization
Information visualization utilizes computer graphics and interaction to assist humans
in solving problems. As such, it incorporates elements of a constructive, formal sci-
ence (the algorithmic and development aspects) with aspects of an empirical science
(for measuring effectiveness and validity). Surrounding both are engineering efforts to
improve the overall system. This section discusses the relationship between these
three parts of the visualization discipline, and suggests that a deeper exploration of
formal, scientific models is needed for a strengthening of the field.
4.1
The Need for Models
Traditionally, visualization has focused on the engineering aspects while importing
“scientific” elements as needed. However, even this borrowing has not been suffi-
ciently utilized. To illustrate this, consider archetypical topics from an information
visualization course syllabus:
Exploration: The process of visual exploration in a larger context
Perception: Fundamental mechanisms for human visual perception
Visual Cognition : How perception translates to thought and action
Color: Aspects of color for visualization
Techniques: Specific visualization metaphors, including interaction
Evaluation: Measuring the effectiveness of a visualization design
A survey of visualization education programs has found that most such programs
focus on visualization techniques (the engineering core) in detriment to the founda-
tional aspects (the scientific core) [19]. As a further example, consider that rainbow
colormaps are still entrenched in visualization research [20] while ample scientific
evidence demonstrates their muddling effect [20,21,22]. The ease of utility for provid-
ing rainbow colormaps does not outweigh the costs in terms of a user's time - the
primary currency of users [23].
Examining the previous list, it is apparent that only the Techniques topic deals ex-
tensively with the engineering aspects of visualization design. While these efforts are
vital to providing actual tools to users, the other elements are needed to provide a
solid foundation to guide those efforts. For example, perceptual literature is grounded
in empirical results with a strong scientific pedigree. A key aspect of these results is
the formal models which are generated to explain the results. Such models are both
descriptive - they encapsulate the factors of the empirical experiment and describe a
mechanism for their operation - and predictive - they generalize the description to a
larger context by predicting future behavior. The predictive nature of the models fa-
cilitates visualization design: it is the predictive nature of color perception models that
explains the limitations of rainbow colormaps [22].
Search WWH ::




Custom Search