Information Technology Reference
In-Depth Information
are processed by separate modules as defined in (Nefti, 2002; Nefti & Djouadi, 2002). This allows us
to consider only eight possible combinations per module except the existence module which has 24
combinations. For the final decision support module, three membership functions are used
Low
,
Aver-
age
and
High
and this gives a possible total of 81 combinations. Table 1 summarizes the information
extracted from the two case studies.
Table 2 shows the fuzzification of the extracted values related to each variable. In the first case study,
the information extraction system found two sub-variables (phone number and address) out of a possible
three; hence a value of 0.6 is assigned to the physical existence variable, thus after the fuzzification
step the degree of membership function of this value is 0.4 for Low and 0.6 for High. The remaining
two variables (people existence and mandatory registration) in the existence factor were assigned the
membership functions 1 (or High) In the remaining case studies, all information related to the existence
factor were found; thus the degree of membership function of all three existence variables is 1. After all
the selected rules were inferred in parallel, the fuzzy operator 'and' is applied to determine the support
degree of the rules. The 'and' results are aggregated and the final trust factor for the case studies were
is generated by defuzzifying the aggregation using the centroid method as shown in Figure 6.
Example of rules used to process the policy_trust_index for the case study 1 is as follows:
IF customer_satisfaction is Low and privacy_statement is High and warranty_policy is low THEN
Policy_trust_index is Low .
IF customer_satisfaction is High and privacy_statement is High and warranty_policy is low THEN
Policy_trust_index is Average.
Table 2. the fuzzification of the extracted information for the case studies
Case Study 3
Case Study 4
Variables
Case Study 1
Case Study 2
Physical Existence
{0.4/L; 0.6/H}
H
H
H
People Existence
H
H
H
L
Registration
H
H
L
H
Affiliation_trust_index
L
H
L
H
Existence_trust_index
H
H
L
H
Endorsement
L
H
H
L
Membership
H
H
L
L
Portal
L
H
H
H
Affiliation_trust_index
L
H
A
L
Customer Satisfaction
{0.2/L;0.8/H}
H
H
H
Privacy Statement
H
L
H
H
Warranty Policy
L
H
A
A
Policy_trust_index
H
A
A
A
Delivery
L
H
H
H
Payment
H
H
H
H
Community Comments
L
L
L
H
Fulfilment_trust_index
L
A
A
H
Search WWH ::
Custom Search