Graphics Reference
In-Depth Information
Graph Changes
One of the most common goals in analyzing dynamic graphs is to be able
to understand how network members and connections change over time or
between different scenarios. For example, looking at the product affinity
data from Chapter 14, “Big Data,” it might be interesting to see the evolution
of interest in associated products over time.
You may recall that, in the affinity example of the
Envisioning Information
book (Cheshire, CT:1990, Graphics Press) from Chapter 14, most associated
products are linked through co-review and that each review has a date,
ranging from Christmas 1998 to the summer of 2001. Narrowing the
window of time and including only those reviews and co-reviewed products
that fall within it will reveal what the neighborhood of associated products
looked like at that time. Scrolling the window of time will lend a taste of how
the market has evolved over the years.
This section examines several methods of visualizing changes between
windows of time in the affinity data set, giving a sense of the relative
strengths of each technique.
Organic Animation
In many ways,
organic animation
(where the beginning and end states
look exactly as they would if treated independently, and the organic
evolution between states is animated) is the most obvious technique for
portraying change in graphs.
Gephi includes a dedicated capability for dynamic graphs where each graph
stateinaseriesofstatesisspecifiedinfullandassociatedwithatime.Atime
slider is then provided to view the graph state over time. If the end goal is to
produce a finished animation of evolution over time, this capability would
be the first step in doing so. However, for the purposes of the exercises
in this chapter, time is better spent exploring simpler and more flexible
approaches to achieve a similar effect.
Return to the subgraph of product interest associated with Tufte's
Envisioning Information
book extracted in Chapter 14. Open it in Gephi,
and select the Filters pane to the right. Filters provide a means of
constraining the set of visible items to those that fulfill specific criteria.
Filters in Gephi can be chained together by assembling them hierarchically.
When assembled in this way, the filter at the lowest level of the hierarchy is
