Information Technology Reference
In-Depth Information
Table 13.2.
Statistical data of virtual user networks month by month. The URL of
this webpage http://bbs.people.com.cn/bbs/. The total number of BBS reply articles
is 745740, those articles posted between Feb. 1, 2003 and Jul. 31, 2003 to form the origi-
nal network. Data preparation process is as follows. Firstly, the articles are downloaded
with multi-thread from the website to storage to HTML files. Secondly, the important
information of an article, such as posting time, poster, reply conditions, are extracted
and deposited to the corresponding database. Lastly, the effective posters are summa-
rized according to the constructed method of article reply network. These nodes are
linked by their reply relationship to form the original network [25]. And in the table,
the effective network is the network with no circle, no multi-edge and removing the
nodes whose original posters are not in the networks.
Month NO
a
EO
b
NE
c
EE
d
NL
e
EF
f
Feb. 3017 75142 1939 9983 1912 9968
Mar. 3936 116940 2401 13132 2375 13118
Apr. 4053 119146 2533 14092 2519 14085
May. 4445 146560 2750 16978 2715 16960
Jun. 5001 141378 2885 15872 2839 15848
Jul. 4497 146574 2625 16278 2572 16250
a
The number of nodes for the original network.
b
The number of edges for the original network.
c
The number of nodes for the effective network.
d
The number of edges for the effective network.
e
The number of nodes for the largest connected component.
f
The number of edges for the largest connected component.
Table 13.3.
The reconstructed networks with different selection threshold
ELR
d
0.0005 447 1752 155(34.7%) 1249(71.3%)
0.0006 710 5608 421(59.3%) 5356(95.5%)
0.0007 1009 18008 744(74.7%) 17836(99.0%)
0.0008 1297 50113 1039(81.2%) 49965(99.7%)
0.0009 1520 107279 1258(82.8%) 107108(99.8%)
0.0010 1725 192621 1647(95.5%) 192442(99.9%)
a
The number of nodes for the reconstructed network.
b
The number of edges for the reconstructed network.
c
The number of nodes for the largest connected component of the reconstructed
network.
d
The number of edges for the largest connected component of the reconstructed
network.
Threshold NR
a
ER
b
NLR
c
overlooked, which forms the more edges. So the nodes and edges of the whole
network are all increasing. Of course, the nodes and edges of the largest con-
nected component also increase.
When these thresholds arrange in sequence from small to large, the theoretical
and actual values of the ratios of the number of nodes of the largest connected
component with the former threshold to that of the latter threshold are shown
in Figure 13.4. For example, when the selected threshold is 0.0005, the number
of nodes the largest connected component is 155, while the latter threshold is
