what-when-how
In Depth Tutorials and Information
folksonomy [16], items represented by corresponding tags can be compared using
several metrics [17,18] such as occurrence of tag pairs:
w t
(
;
t
) :
=
card u r U R t
{( ; )
∈ ×
|
;
t
∈
T
ur
}.
1
2
1
2
: ( ; ; ;
=
where
U
,
T
, and
R
are finite sets of
users, tags, and resources respectively.
Y
is a ternary relation among them,
i.e.,
Y U T R
(A folksonomy is a tuple
F
U T R Y
⊆ ×
×
.
A post is a triple
( ;
with
u
∈
,
r
∈
, and
u T r
; )
ur
T
ur
:
= ∈ ∈
.
he similarity between
t
and
�
{
t T u t r
| ( ; ; )
Y
}
t
is represented as follows:
k
n
∑
1
ʹ =
ʹ
=
ʹ
≥
0
sim t
( ,
t
)
sim t
(
, )
t
w t
( ,
t
)
i
j
=
=
i
1
j
{
t
n
}
=
ʹ
champ
max
sim t
( ,
t
),
sim t
( ,
t
),...,
sim t
(
,
) .
1
hus, the privacy policy associated with
champ
is prompted to all the users in the
same set.
Disadvantages:
he approach proposed in Reference 14 should be extended for
a hierarchical relationship structure. Besides, it is necessary to take the type of
access privilege into account. In order to determine the optimal privacy policy and
calculate the tax, Reference 14 requires every user to publish their social value
v g
i
( )
simultaneously. However, this requirement, which needs an effective syn-
chronization mechanism, may not be achieved in distributed networks. he Clark
Tax approach is not perfect because of the assumption that users should be able to
calculate the value of diferent privacy policies. his assumption cannot be achieved
easily. Fortunately, the approach is implemented with short execution time.
10.5. PrivacyPreservinginCollaborativeSTNs
Reference 19 proposes a method to preserve the privacy while extending the indi-
vidual STN to multiple collaborative STNs. his kind of collaborative network
must address three problems as follows:
10.5.1 Privacy Problem 1
Combine the multiple and sometimes conflicting data from different providers in
different STNs into a single network.
his step not only is a mere extension of the existing algorithm for creating the
network but also should concern about collaborations among STNs:
Criteria.1
: Given the collaborative STN
S,
which contains
n
social networks,
n
1
cannot determine the contents of
n
2
when
n
↑
.
n
1
