Graphics Reference
In-Depth Information
Identifying these groups of people and how they are connected can help a
company identify different customer segments and better understand the
dynamics of influence within and between them.
In a large company, social network problems may easily involve millions
of nodes. Representing these graphs visually and exploring them for the
purposes of extracting meaningful information is exceptionally difficult.
Common desktop tools like Gephi (which are limited by in memory
processing on a single machine) are not designed for graphs of that size.
We are involved in an ongoing advanced research effort exploring the use of
cluster computing for community-detection and graph-drawing techniques
to achieve highly scalable zoomable graphs with millions of nodes and tens
of millions of links.
Figure 1-15
shows an example of one such graph
involving referrals. Clusters of medical practitioners seeing the same
patients are outlined with circles, indicating communities.
