Information Technology Reference
In-Depth Information
for each user v, u
V do
update value of N
vu
end for
for each user v, u
∈
∈
V do
N
/
N
P
vu
⇐
vu
vx
xoN
∈
()
update value of P
vu
end for
i
⇐
1, BR
1
⇐
AR, ε
⇐
0.0001, σ
⇐
1
while σ ≥ ε do
BR
i+1
⇐
(1 − d)
×
×
×
BR
i
+ d
P
BR
i
σ
⇐
|| BR
i+1
− BR
i
||
i
⇐
i + 1
end while
return B R
4.3
To Establish the Model of IR
The first two chapter studied effects on influence of nodes from aspect of node
attributes and interaction behavior. According to the above algorithm, the two aspects
will be integrated. We use them as nodes influence factor synthesis in online social
networks influence.
After a lot of observation and analysis: node attribute is a basic condition
determines the influence of nodes, if the node attribute value AR is weak, the node will
not influence overall strong. At the same time, the interactive behavior between the
nodes can also change the node influence; this kind of behavior is equivalent to the
influence in the network spread, similar to the random walk model. The interaction
behavior between nodes and the attributes of the nodes combined in accordance with
the appropriate weight, a formula to calculate the influence of nodes:
IR i
()
=
α
AR i
()
+
(1
−
α
)
BR i
()
i
=
1, 2,
(10)
,
n
α
In formula (10)
is regulatory factor, it can regulate the weight of AR and BR. This
formula is a linear combination of AR and BR. When
0.5
<<
1
, it is shown that the
node attribute is more importance. When
0
<<
0.5
, it shown that we more
emphasis on interactive behavior between nodes to measure the influence.
5
Experiments
The paper selects the Sina micro-blog as the data source, the data obtained by Sina
open API.The data set is part of Sina micro-blog user information in December 2012.
It contains 60290 users. Every user includes a user ID, user type, numbers of fans,
numbers of micro-blogs, numbers of attention, forwarding numbers, the number of
comments.
