what-when-how
In Depth Tutorials and Information
he SocialTrust model can be optimized to balance the influence among the three
components by tuning the parameters
α
,
β,
and
γ
.
11.4.3 Indirect/Direct Trust Models
he models mentioned above are used to calculate the global reputation of the
target based on the assumption that all information is available (there must be a
centralized trust management center that assembles and processes the trust infor-
mation) and every node has the same trust value towards the target. Possibly this
assumption is unrealistic in the distributed and complicated environment of the
STNs, for example,
x
may trust
z
while
y
does not.
hus, we need to establish a distributed trust management system between
source and target in STNs. he overall trust consists of direct trust and indirect
trust [16] that are reflected by the recommendation obtained from the other
nodes' transactions with the recommended node. A source can query other
nodes to get the recommendation of the target according to the characteristics
of STNs.
For example, in Reference 16 direct trust is denoted as
Suc
N
ij
=
dr
(11.4)
ij
ij
where
Suc
ij
denotes the number of successful transactions between node
i
and node
j
.
N
ij
denotes the total number of transactions.
T
ij
denotes the value of recommen-
dation from
i
to
j
.
−
−
Suc
(
N
Suc
)
ij
ij
ij
=
T
∑
ij
(11.5)
Suc
ik
k
Indirect trust is denoted as
k
∑
⋅
+
λ
R T
(
)
ij
ij
(11.6)
=
ir
=
i
1
ij
k
where
R
ij
denotes reputation from
i
to
j
. It decreases exponentially due to the pun-
ishment for the false recommendation and increases linearly as the recommenda-
tion is accurate. he variable
λ
denotes the adjustment coefficient to ensure that the
network is stable. hus, the overall trust is
= ⋅
α
+ −
α
⋅
Trust
dr
(
1
)
ir
ij
ij
ij
(11.7)