what-when-how
In Depth Tutorials and Information
user's behavior will associate with the group's behavior. Values of 3, 4, and 5 are
appropriate but, typically, value 3 is used due to its appeal at reducing algorithm
complexity. he inal experience parameter,
λ
, correlates the decay rate of the user's
interest with the elapsed time. his value can be changed between 0.3, 0.6, and 0.9,
with 0.3 projecting the curve above the real data, 0.6 coinciding with the real data,
and 0.9 projecting the curve under the real data. his is useful in inding the lower
and upper bounds of the equation data.
In conclusion, probability models are very useful in the modeling of social net-
works. Using probability, predictions can be made accurately, and entire network
structures can be defined simply by one or two equations. Since there is a need for
predictability in network modeling, probabilistic models are highly advantageous
to the development of an experimental social network model design.
5.3 DynamicSocialNetworks[3]
Social networks are highly different from other types of networks in that they are
extremely dynamic. A social network structure is constantly changing and updat-
ing due to the human and social element involved. Most analyses of social networks
have involved static computations, but adapting mathematical and computational
frameworks that involve dynamic aspects are becoming more popular.
In the standard static network models, time is essentially discarded. his
static nature can introduce inaccuracies and inexact data regarding data pat-
terns [2]. his is because many patterns of dynamic data can be used to form the
same static results, which is inconclusive and incorrect. Static models also prevent
information about cause and effect and consequences from being recorded and
analyzed.
he main concept defined to represent a dynamic network model is a meta-
group. An individual user is a member of a given meta-group if the number of
groups to which that user belongs is at least an
apriori
chosen membership thresh-
old function. he following two rules must be adhered to:
◾
No two groups in the meta-group can be in the same partition, and the groups
are ordered by partition time steps.
◾
he consecutive groups in the meta-group are similar due to a certain func-
tion and parameters.
hree values,
α
(persistence),
β
(turnover), and
γ
(membership), give the meaning of
a group. he framework is independent of the deinitions and capable of providing
significant information and data. Using a weighted multipartite directed acyclic
graph, the conceptual representation is made. he graph is acyclic since all edges
move forward to another point in time and never backward or sideways. he graph
is called a meta-group
β
-graph.
