what-when-how
In Depth Tutorials and Information
G2
G3
a3
a5
Action()
JOIN
a2
a4
a1
G1
Figure 5.1
A probabilistic social group evolution model; before JOIN action.
(From H.-C. Chen et al., Personalization inferring agent dynamics from social
communicationnetworks,in
International Conference on Knowledge Discovery
and Data Mining
,2007,pp.36-45.)
heories of probability can be used with evolution models as well. In the exam-
ple shown in Figure 5.1, there are five actors and three social groups involved with
the model. his probabilistic social group evolution model shows that, based on
the properties of itself and other actors,
α
1
decides to join a new group through the
stochastic process shown as
Action
() in Figure 5.1. his stochastic process depends
on a set of parameters
θ
action
, with other possible actions including leaving a group
and doing nothing.
To find the information on which group to join,
α
1
must gather information
through its neighbors
α
2
and
α
4
. his information includes which groups the neigh-
bors belong to. Based on these references,
α
1
can infer that the possible groups to
join are
G
2
and
G
3
in the case shown in Figure 5.1. Based on its own actor qualifi-
cations,
α
1
then decides which group to apply to. his decision weighs the actor's
qualifications as measured by the average rank (by all groups) and qualification
thresholds and sizes of the potential joining groups.
Once
α
1
decides to join a group, it must apply to the specific group it has
decided to join. his can be accomplished through a stochastic handshaking pro-
cess. First,
α
1
decides to “apply” to group
G
2
. At this point,
G
2
decides whether or
not to accept
α
1
as a part of its group. his process is governed by a set of parameters
θ
group
and is depicted in Figure 5.2 as
Group
(). his process is similar to the one
used to decide which group the actor wants to join, and the same actor qualifica-
tions are used in the group's decision.
Another example of using a probabilistic approach to describing social network
models is presented in Reference 2. In considering an online social network, a dynamic
probability model can be used, and this model is stated in three hypotheses.
he first hypothesis states that, through the individual interest factor, it can
be assumed that the more times one user attends a discussion about a topic T, the
