Information Technology Reference
In-Depth Information
quality measured), the
Y
axis should start from 0, as starting from a higher value can
exaggerate degrees of change or difference. If the axis is started at a higher value for
clarity, the reader should be alerted to this in the caption.
Beware of using graphs to make unsupported claims. For example, consider the
line labelled “space wastage” in the graph in Fig.
11.3
: it would not be possible to
identify the slope of this line with any confidence, nor identify it as a particular kind
of curve. The only reasonable inference would be that increasing list length generally
increases space wastage.
There are many good software packages for drawing graphs. Valuable features
include:
Placing of several lines on one graph.
A range of symbols (such as crosses, squares, and triangles) for marking points.
The ability to create custom symbols of custom size for marking points.
Optional connection of points with solid, dotted, or dashed lines, and optional
omission of the point marks.
The ability to place text at specified places in the graph.
Multiple font sizes and line thicknesses.
The ability to use the same fonts as in the body of the text.
Availability of greys and colours.
Optional logarithmic or exponential scaling on both axes.
Axis editing, to specify where the ticks are placed, how many digits of precision
to use, and what range to cover.
The ability to move and rotate the legend or key, line labels, axis labels, and the
graph label.
The ability to apply simple functions or external programs to
(
x
,
y
)
values.
The ability to lay out mathematical symbols and basic expressions.
The generation of images in vector format (postscript, PDF, SVG) rather than
raster format (jpg, gif, png), to allow subsequent rescaling.
Most of these features were used in the example graphs in this chapter.
Graphs and diagrams attract the attention of readers, so should be reserved for
material that is central to the paper.
Diagrams
Diagrams serve many purposes in computing papers. They illustrate processes or
architectures, explain data structures and algorithms, present relationships, visual-
ize data, and show examples of interfaces. There are areas of computer science in
which the diagrams are, in some sense, the result being presented in the paper:
entity-relationship models are diagrams conforming to a well-defined notation, for
example, and automata are often described by diagrams. A novel visualization of a
massive dataset can be a potent demonstration of previously unknown properties or
behaviour.
