Information Technology Reference
In-Depth Information
Let us now introduce some formal constructs. We define
Act
=
{
α
1
,..,
α
n
}
be a finite set of
actions
, and
Agt
a finite set of
agents
. Each agent has an action repertoire, a
plan library, resources, goals, beliefs, motives, etc.
As introduced in Chapter 2, the action/goal pair
=
{
X, Y, A, B,..
}
τ
=
(
α
,g)
is the real object of delegation,
and we called it 'task'.
1
Then by means of
τ
, we will refer to the action (
α
), to its resulting
world state (
g
)
2
, or to both.
Given an agent
Y
and a situational context
(a set of propositions describing a state of
the world), we define as trustworthiness of
Y
about
τ
in
(called
trustworthiness (Y,
τ
,
.This
objective probability is in terms of our model computed on the basis of some more elementary
components:
)
), the objective probability that
Y
will successfully execute the task
τ
in context
An
objective degree of ability
(
OdA
, ranging between 0 and 1, indicating the level
of
); we can say that it could be measured as the
number of
Y
's successes (
s
) on the number of
Y
's attempts (
a
):
s/a
, when
a
goes
to
Y
's
ability
about
the
task
τ
∞
:
OdA
Y
=
lim
(
a
->
∞
)
s
/
a
(6.1)
and
An
objective degree of willingness
(
OdW
, ranging between 0 and 1, indicating the level of
Y
's intentionality/persistence about the task
); we can say that it could be measured as the
number of
Y
's (successfully or unsuccessfully) performances (
p
) of that given task on the
number of times
Y
declares to have the intention (
i
) to perform that task:
p/i
, when
i
goes to
∞
τ
:
OdW
Y
=
lim
(
i
->
∞
)
p
/
i
(6.2)
we are considering that an agent declares its intention each time it has got one.
So, in this model we have that:
Trustworthiness
(
Y
,τ,
)
=
F
(
OdA
Y
τ
,
OdW
Y
τ
)
(6.3)
Where
F
is in general a function that preserves monotonicity, and ranges in (0,1): for the
purpose of this work it is not relevant to analyze the various possible models of the function
F
. We have considered this probability as
objective
(absolute, not from the perspective of
another agent) because we hypothesize that it measures the real value of
Y
's trustworthiness;
for example, if
trustworthiness(Y
τ
)
=
0.80
, we suppose that in a context
,
80%
of times
Y
tries and succeeds in executing
.
As the reader can see, we have considered the opportunity dimension as included in
τ
:the
external conditions favoring, allowing or inhibiting, impeding the realization of the task.
1
We assume that
to delegate an action necessarily implies delegating some result of that action.
Conversely,
to
delegate a goal state always implies the delegation of at least one action (possibly unknown to Y) that produces such
a goal state as result.
2
We consider
g
=
g
X
=
p
(see Chapter 2, Section 2.1).
Search WWH ::
Custom Search