Information Technology Reference
In-Depth Information
negative beliefs is more complex. For this it is necessary a more subtle (fine-grained) analysis
of features and properties.
Let us now try to refine the previous analysis. Starting from our characterization of the
tasks, we can say that the similarity between different tasks is given by the overlap among
the properties, in particular among their core properties (the indispensable relevant ones). So,
the interesting thing is that, if the trustor believes that the trustee has the right features for a
specific task, and another task has a significant overlap of its properties with that task (and
specifically they share the core properties), then the trustor could also believe that the trustee
has the right features for realizing the second task.
Also, in general, we can say that, given two tasks
τ where both the formulas (6.17)
and (6.17bis) are true, if the trustee (on the point of view of the trustor) has the right features
for realizing
τ
and
τ
and:
τ C
C
τ
(6.22)
then, on the trustor's point of view, that trustee also has the right features for realizing the
task
τ .
So, even if the trustor had never seen the trustee operate on the task
τ , it can infer its
trustworthiness on that task. Note that the trustor could trust the trustee on a new task
τ also in
the case in which they ignoring their features, but has experienced their trustworthiness about
τ
(in fact deducing them from the previous task in an indirect way).
In the case in which:
τ C
C
τ
(6.23)
there is at least one p j
τ C ) that is different from all the p j
C ) . It means that in this case
(
(
τ
the trustor can trust the trustee if: for each of the p j
τ C ) either it is equal to one of the p j
(
C ) , or the trustor believes that it is satisfied from at least one of the trustee features ( f AgY
(
τ
τ if they believe
f 1 ,
...
,f n }
). In other words, the trustor can trust the Ag Y on a different task
{
that:
either all the core properties of τ
are included in the core properties of
τ
(where Ag Y is
);
or for each property of τ not included in τ Ag X believes there is at least one feature of Ag Y
that satisfies that property (and, of course, Ag Y is believed to be trustworthy on the task
believed trustworthy on the task
τ
τ
).
Of course, all the considerations made in this paragraph can also be applied to self-confidence
and self-efficacy; Ag X 's trust about himself: 'Given Ag X 's success on task
τ
will Ag X be able
τ ?'; 'Given Ag Y 's success on
τ too?'.
on task
τ
will Ag X be able on
6.6.5 Classes of Agents and Tasks
It is quite simple to extend the above reasoning to a class of tasks. We can define a class of
tasks as the set of tasks sharing a precise set of (quite abstract) core properties. The different
specifications of these properties define the different tasks belonging to the class. On the basis
of the core properties of a class, we can select all the agents endowed with the appropriate
features who will be trustworthy for the whole class of tasks.
Search WWH ::




Custom Search