Graphics Reference
In-Depth Information
In the age of social media, understanding key social community influencers
can be important to building and maintaining a positive company image
and reputation for many businesses. Knowing which opinions matter most,
and to which social groups, enables you to focus and tailor your marketing
efforts and determine when rapid action is required in response to
developing negative chatter.
You can also direct community analysis inward on your own performance to
revealpatternsinthesuccessorfailureofbusinessinitiatives.Likeamedical
scientist who seeks to identify factors behind illnesses and to understand
how they affect various communities of people, you can use insights into
communities of company projects to identify key performance drivers and
institute changes that will positively affect the health of the business.
This chapter begins with a definition of community, introduces graph
clustering, and steps through an example social media application
beginning with the ingestion of data. Layout, color, and filtering techniques
for visualization are then described. Aspects of community detection
algorithms are defined, and one is chosen for the exercise. The example
continues with community topic analysis, followed by an introduction to
cliques and use of convex hulls. By the end of this chapter, you should be
familiar with basic use of NodeXL and Gephi, as well as key techniques for
visualizing and analyzing community structure.
What Defines a Community?
In graph data, a
community
is a cluster of nodes with a relatively high
density of internal connections, as shown in
Figure 11-2
.
In technical terms,
it is a set of nodes with high
modularity
. Communities may overlap each
other. They may also be nested, such that higher-level communities are
formed from more localized communities.
Visualization can be indispensable when trying to understand the nature of
identified communities and their collective structure. Their makeup, their
relative distance, and how they overlap (and to what degree) are all aspects
that are difficult to describe in words, but that are naturally expressed
visually.
Graph community visualization works similarly to many kinds of geospatial
visualization in that the spatial coordinates of a node or data point are not
important at the individual level, but rather in how they visually resolve into
