Information Technology Reference
In-Depth Information
system automatically assigns the value of 1 when a variable is found and 0 otherwise. The total is then
divided by the number of variables (12) and a trust factor with a value in the interval [0,1] is calculated us-
ing equation (4) were
T
is the trust factor and
v
i
.represents one of the 12 variables of the trust model.
1
12
12
∑
(4)
T
=
v
i
i
=
1
The parameterised model is used with more experiences users which are asked to evaluate the
importance (according to their perception) of each variable by assigning the value 1 if the variable is
judged important, 0.5 if the variable is fairly important and 0 if it is not important. These values are
used as weights to the linear model variables and again a trust factor
T
in the interval [0,1] is calculated
using equation (5).
3
3
3
3
∑
A
w
∑
E
w
∑
P
w
∑
F
w
i
1
4
i
i
i
i
i
i
i
j
=
1
T
=
I
=
1
+
+
k
=
1
+
l
=
1
(5)
3
3
3
3
∑
∑
∑
∑
w
w
w
w
i
i
i
i
i
=
1
j
=
1
k
=
1
l
=
1
In both cases, the closer to 1 the trust factor is the higher the trust is towards the merchant's website.
We have compared the results obtained by the Fuzzy model with those obtained by the linear and pa-
rameterised model and the results are shown in Table 4. The weights used for the parameterised model
used in this experiment, chosen by the authors, are as follows: Physical existence (1), People existence
(0), Mandatory registration (1), Third Party endorsement (1), Membership (0.5), Portal (0.5), Customer
Satisfaction (1), Privacy statement (1),Warranty (1), Delivery (0.5), Payment (0), Community comments
(0.5).
The results obtained by the linear model, which are those provided for a new or inexperienced user,
are high compared to those obtained by the other two models. The linear model results are only based
on the existence or non existence of the variables on the merchant's Website and this may be misleading
as not all variables are of the same importance. The results of the fuzzy and parameterized models are
close because in this particular experiment, those who choose the weights for the parameterized model,
namely the authors, are themselves experts and have experience in EC transactions. Hence, the results
are similar to those produced by the fuzzy systems where experts have produced the IF-THEN rules.
The small differences can be justified as the aggregation method used in the parameterised model
given in equation 5, do not capture the notion of uncertainty as the weights are only limited to three
Table 4. Trust models comparison
Fuzzy Model
Linear Model
Parameterised Model
Case study 1
0.62
0.66
0.60
Case study 2
0.76
0.83
0.79
Case study 3
0.5
0.75
0.5
Case study 4
0.58
0.75
0.54
Search WWH ::
Custom Search