what-when-how
In Depth Tutorials and Information
networks [43]. Besides, through modeling and analyzing user behaviors and
relationship between users, we can discover the important users [27-30] who
play key roles in the information propagation process, design an effective propa-
ganda strategy to maximize the spread of information through a social network
[65, 68], mine some latent business opportunities for corporations [39], and an
effective information-recommending mechanism to supply suitable information
to each type of users [52, 63]. In this section, we focus on the recent research
progress on the analysis and models of user behaviors in the information propa-
gation process.
3.2.1 Relationship between Users
As one of the most significant characteristics of online social networks, the rela-
tionships between users mainly include blogroll link [18, 19, 22, 25], information
interaction [20, 23, 24], and the connection between social networks and outside
information networks [21, 23], etc. To analyze the information propagation pro-
cess, we should understand the relationship between users and find out how the
relationship influences the process.
Blogroll links are usually located in the blog's sidebar and point to other blogs that
the author may read or that may simply be the author's friends' blog. Blogroll links are
the most simple and common relationship between users. Prior works have investi-
gated the quantitative features of blogroll links, such as overlap, distance, reciprocity,
in-degree and out-degree, density and centralization, community structure, etc.
3.2.1.1 Overlap
Overlap is defined as the intersection between the sets of nodes that are reachable
from each starting point. Considering all the paths having three hops for blog
a
,
blog
b
, and blog
c
, and four hops for blog
d
, then classifying all the nodes reached
into sets that were reached by
a
,
b
,
c
, or
d
individually, or in any combination, the
members of these sets were classified as A-list [19] or not and counted. A chi-square
test for independence was conducted. he test is signiicant (chi-sq = 194.2, 3 df,
p
< .0001), suggesting a strong association between a blog's membership in the
A-list and how reachable it is from the four starting blogs. he largest deviance
(177.2) came from A-list nodes reachable from all four starting points, suggesting
that A-list blogs tend to be reachable from any starting point, whereas the same is
not true for non-A-list blogs. his result indicates that A-list blogs are more central
in the network than other blogs.
3.2.1.2 Distance
Distance is defined as the shortest paths between users. Mislove Alan et al. [25]
analyzed the data of four online social networks (Flickr, LiveJournal, Orkut, and
