Information Technology Reference
In-Depth Information
Firstly, we construct matrix of the influence factor, and utilize analysis hierarchy
process to determine the weight of each node, and then calculate node's
AR
.According
to the network relationship of micro-blog, and the matrix of transition probability, we
can get the node's
BR
.Finally, and we calculate the node of the
IR
value through
Influence Rank model proposed in this paper.
Table 2.
The Calculation Results of Top 10 Users
UserID
AR
BR
IR
1266321801
1.0000
0.4409
0.6086
0.4590
1192329374
0.7045
0.3538
1656809190
1087770692
1192515960
1752467960
1742727537
1730336902
1212812142
1854283601
0.5305
0.5508
0.3304
0.3249
0.1000
0.1331
0.2137
0.0601
0.4265
0.3814
0.3977
0.3390
0.4233
0.4046
0.3683
0.4126
0.4577
0.4322
0.3775
0.3347
0.3263
0.3231
0.3219
0.3069
Because the data set is large, we list the influence of top ten users. In the process of
calculation, we assumed
α
is 0.3.Node's
IR
value is not only closely related with the
AR
and
BR
values, but also with the relevant to the value of
α
. In general, if the
influence of node is greater, the
AR
and
BR
will be greater.
Fig. 2.
Comparison of different IR
α
taking different values, the results of IR are different.
As is shown in Figure2 the node's IR value is closely related to regulatory factor
Like formula (10), when
α
.
The regulatory factor represents the weight of AR.As is shown in TABLEII, the 10
th
node's AR is very small, when
α
is larger the IR will be smaller as shown in Figure2.
α
Therefore,
is used to balance of AR and BR according to need.
