A Decision Support System for Trust Formalization - Distributed Artificial Intelligence, Agent Technology, and Collaborative Applications - page 53

where t-norm ⊗ and s-norm ⊕ are used for the intersection and the union operations respectively. The
final output is a fuzzy set in V , which is a combination of the M fuzzy sets, A ′ ° ( R (1) ,..., R ( M ) ). The mem-
bership function of this inferred fuzzy set will be:

(3)

( ) y = ( ) y (1) ( ) M (1) ( ) M

A R ' (  ,..., R ) A R '  ⊕ ⊕ ... A R ' 

B y . The crisp value
of the output action can be obtained, say, by using the Centre of Gravity (COG) defuzzification method,
where the shape of membership function,

The above membership function defines the fuzzy value of the output action ( )

, is considered to determine the crisp value of

( ) y (1) ( ) M

A R ' (  ,..., R )

the output action

y = ∑
∑

y

.

B

'

B

Justifica tion of the use of fu ZZY Log IC To eVALuATe Tru ST

Trust relationships among customers and vendors are hard to assess due to the uncertainties and ambiguities
involved in evaluating trust in EC. For example, in the proposed trust model, the community comments
variable in the fulfilment factor has a wide range of values as we may have a small or a large number of
customers providing positive or negative feedback to the vendor. Hence, the number of comments and
their nature will affect the decision made by the associated evaluation module. In addition, in the trust
model used, there are dependencies between some variables. For example the mandatory registration
variable in the existence factor is dependent on the membership and third-party endorsements variables
in the affiliation factor. Indeed, if an organization is a member of an association or endorsed by a third
party, we assume that this organization is fully registered with the required authorities even though the
mandatory registration was not extracted by the information extraction system.
Thus, the use of fuzzy reasoning is justified as an adequate approach to deal with evaluating trust
in EC as it has the ability to quantify imprecise data and quantify uncertainties in measuring the trust

Figure 3. The ecommerce fuzzy trust model