Mutual information to assess structural properties in dynamic networks

This
 article
 proposes
 applying
 the
 variation
 of
 information
 measure
 from
 Information
 Theory
to
evaluate
macro‐level
properties
characterising
dynamic
networks.
This
measure
is
used
to
 evaluate
 different
 clusters
 given
 by
 the
 agglomerative
 hierarchical
 clustering
 algorithm
of
 Clauset,
 Newman
and
Moore
 (2004),
 concerning
a
 case
 study
of
 the
multi‐agent
based
network
model
of
 a
 university
email
service.
The
variation
of
information
measure
is
shown
to
be
capable
of
assessing
the
 outcome
of
simulating
the
dynamics
of
networks,
in
terms
of
its
macro‐level
properties.
 Keywords:
 social
 networks,
 community
 detection,
 clustering
 algorithms,
 multi­agent
 simulation,
 information
theory,
email