Graphics Reference
In-Depth Information
neighborhoods and geopolitical structures in the aggregate. It is, however,
fundamentally different in that those neighborhoods and communities
cannot be identified by their locations. Because spatial location in a graph is
abstract,communitycharacteristicsmustbevisuallyexpressedinadifferent
way.
Figure 11-2: Graph communities are sets of densely connected nodes,
illustrated here by color.
There is no one precise mathematical definition of a community, nor is
there one single approach for detecting one. The suitability of a technique
depends on the nature of the data and the problem you are trying solve,
and algorithms must often be tuned in individual cases for best results. The
following two sections illustrate two common ways to look at community,
clusters, and cliques, using the same case study data set.
Graph Clustering
The most common computational approach to community detection is to
use graph clustering. Clustering is a goal-driven process by which sets of
similar elements are grouped. Algorithms for achieving those goals vary, as
do the goals themselves. One goal may be to organize data into a certain
 
Search WWH ::




Custom Search