Database Reference
In-Depth Information
contribution. If the algorithm itself is straightforward enough that only a brief
description is required, or if all of the techniques that you use have been ade-
quately described in previous work, then you would be better served by explicitly
writing a Design Study paper.
2.3 Design Study
Design Study papers make a case that a new visual representation is a suitable
solution for a particular domain problem. First, you should explain the target
problem. You must provide enough background that the reader can pass judge-
ment about whether your solution is good, but not so much detail that the focus
of the paper is on domain problems rather than infovis issues. Finding the right
balance is a dicult but crucial judgement call. Second, you should crisply state
the design requirements that you have determined through your task analysis.
Third, you should present your visual encoding and interaction mechanisms and
justify these design choices in terms of how well they fulfill the requirements.
Typical arguments would refer to perceptual principles and infovis theory. For
example, using spatial position to encode the most important variables and using
greyscale value rather than hue to encode an ordered variable are both very de-
fensible choices [24]. The best justifications explicitly discuss particular choices
in the context of several possible alternatives.
Fourth, you should present results that back up the claim that your approach
is better than others. Typical results include case studies or scenarios of use. De-
sign studies often document iterative design and the use of formative evaluation
for refinement. The research contribution of a design study is not typically a new
algorithm or technique, but rather a well-reasoned justification of how existing
techniques can be usefully combined. For most design studies, adoption by the
target users is valuable evidence that the system has met its goals, as are anec-
dotes of insights found with the new system that would be dicult to obtain
using previous methods.
I think this category name is still a good choice, despite the fact that great
design studies are all too rare. I argue that the field would be well served if more
authors explicitly cast their work in this category. Interesting examples that ap-
proach the design study from several angles are the cluster-calendar system [40],
ThemeRiver [11], Vizster [13], VistaChrom [21], and a hotel visitation pattern
analysis system [42].
2.4 Systems
Systems papers focus on the architectural choices made in the design of an in-
frastructure, framework, or toolkit. A systems paper typically does not introduce
new techniques or algorithms. A systems paper also does not introduce a new
design for an application that solves a specific problem; that would be a de-
sign study. The research contribution of a systems paper is the discussion of
architectural design choices and abstractions in a framework or library, not just
a single application. A good example is the prefuse systems paper [14], which
Search WWH ::
Custom Search