Information Technology Reference
In-Depth Information
introduce the socio-cognitive approaches, which have been discussed at length in the other
chapters. In addition you can refer to Chapter 11 for a description of a fuzzy implementation
model for the socio-cognitive approach to trust.
12.2.1 Logical Approaches
The logical approaches ((Cohen and Levesque, 1990), (Fagin e tal. , 1994), (Demolombe,
1999), (Jones and Firozabadi, 2001), (Josang, 2001), (Liau, 2003), (Lorini and Demolombe,
2008), (Castelfranchi et al. , 2009)) start from models based on mathematical logics for describ-
ing, analyzing and implementing the trust relationships. These approaches have the advantage
of using powerful methods able to produce inferences and strongly rationalize the conceptual
apparatus. The drawbacks are given as the constraints introduced and derived through the same
logics that in fact impose their own rules to the conceptualization of thought and action, very
often without considering all the elementary criteria of flexibility of the human reasoning and
action. In various cases, the approximation of the formalisms to the reality can satisfy specific
descriptive purposes of the reality, in other cases this approximation is not appropriate (in the
sense that it does not introduce realistic constraints).
A very elegant example of logical approach is given by the Demolombe's analysis
(Demolombe, 1999) with respect to the trust in information sources. With a clear (but func-
tional) simplification (maybe too superficial with respect to the social concepts he intends to
model), he considers that each information source can have four different properties: sincerity ,
credibility , cooperativity , and vigilance .
An agent X is sincere with respect to another agent Y and a specific content p if X believes
p when he is informing Y about p .
An agent X is credible with respect to another agent Y and a content p if X believes p and p
is true in the world.
An agent X is cooperative with respect to another agent Y if what he believes is communicated
to Y .
An agent X is vigilant with respect to the world if what is true in the world is believed by X .
Demolombe also derives two other concepts: Validity , as the conjunction of sincerity and
credibility; and Completeness , as the conjunction of cooperativity and vigilance.
Using the modal logic (Chellas, 1990) Demolombe formalizes these concepts and defines
different kinds of trust: trust with respect to sincerity, credibility, cooperativity, and vigilance.
Then he is able to derive consequences from this representation: For example, he is able to
derive additional properties like 'if Y trusts X as sincere, then the information p (received by
X )letinferto Y that X believes p ', and so on.
Jones (Jones and Firozabadi, 2001) represents some interesting aspects of trust (like de-
ception in terms of trust, trust in other's trust) applying the logic of belief together with the
deontic logic, and with the logic of 'count as'. We deeply analyze the Jones's model in another
part of this topic (see Chapters 2 and 8 and in particular Section 2.2.2).
One of the main problems of applying the classical logical framework to the mental attitudes
(on which trust is based) is the difficulty of taking into consideration uncertainty and ignorance :
typical features of the beliefs. Different and interesting attempts (in particular the one of
Search WWH ::




Custom Search