what-when-how
In Depth Tutorials and Information
several definitions of interest similarity, with which users' activities could be pre-
dicted. Cheng et al. [35] proposed a bloggers' interests modeling approach based
on the forgetting mechanism. Short-Term Interest Models (STIM) and Long-
Term Interest Models (LTIM) are constructed to describe bloggers' short-term
and long-term interests. Experiments show that both models can identify blog-
gers' preferences well.
3.3 ModelsandAnalysisofInformationFlow
In this section we introduce prior research on models and analysis of information
low in online social networks, including analysis of the information low pathway
[53, 55, 57] and models of innovation diffusion [54, 56, 58-62, 64]. It is different
from the research on user behaviors in that this research focuses on the dynamic
propagation process on a larger scale.
3.3.1 Discovering and Analyzing the
Information Flow Pathway
In blogspace, each information low pathway indicates a sequence of blog users who
post articles related to the same topic sequentially. he pathway can identify the
users who discussed the same topic and the sequence of these users.
Based on a closed-world assumption (that in a given blog community all posts
on a topic except the first one are the result of communication within the com-
munity), Stewart et al. [53] defined the problem of discovering the information
propagation pathway from blogspace as a frequent pattern mining problem. he
following are a few necessary definitions:
Deinition 1 . [Blog community] A blog community collected in a given time
period [
is a set of n blog Ω = {
= (
1 2 con-
tains a set of published posts, where each post is associated with a publishing time
point, T p i
t
,
t
]
b b
,
,...,
b n
}
. Each blog b
p p
,
,...,
p m
)
s
e
1
2
( ) .
Deinition 2. [Topic blog sequence] Given a particular topic c , a topic blog
sequence Q c
(
) such that t
ʺ
T p
ʺ
t
s
i
e
( )
= <
(
b t
,
),(
b t
,
),...(
b t
,
)
>
1 1 2 2 is a list of blog-time pairs such that
each blog b i publishes a post on the topic c at time t i . Moreover, ∀ ∈
k
k
[ , ),1 1 .
Deinition 3 . [Blog sequence database] Given a blog community collected
in time period [
k t
+
t
i
i
i
t
,
t
]
, which contains a set of blogs Ω = {
b b
,
,...,
b n
}
and a set
s
e
1
2
of posts on k topics Γ = {
c
,
c
,...,
c k
}
, it can be modeled as a blog sequence data-
1
2
base D in the form of ( ,
i Q c i
(
))
, where i
(
1 ʺ ʺ is the identity of a topic  and
i
k
)
= <
>
Q c
1 1 2 2 ( c i ∈Γ , b i ∈Ω, t s t i t e ) is a topic blog sequence.
Deinition4 . [Support] Given a blog sequence database D and a blog sequence
S , the support of S with respect to D , denoted as SuppD , is the fraction of the topic
blog sequences in D that support S .
(
)
(
b t
,
),(
b t
,
),...(
b t
,
)
i
m m
Search WWH ::




Custom Search