Information Technology Reference
In-Depth Information
1 : requests(?
team,
?
goal
)
∧
isPartOf(?
agent
1
,
?
team
)
∧
isPartOf(?
agent
2
,
?
team
)
∧
2 : hasSubgoal(?
goal,
?
g
1)
∧
hasSubgoal(?
goal,
?
g
2)
∧
3 : hasDelegation(?
team,
?
d
1)
∧
hasDelegatee(?
d
1
,
?
agent
1)
∧
hasDelegatum(?
d
1
,
?
g
1)
∧
4 : hasDelegation(?
team,
?
d
2)
∧
hasDelegatee(?
d
2
,
?
agent
2)
∧
hasDelegatum(?
d
2
,
?
g
2)
∧
5 : hasNegDContribution(?
goal
1
,
?
goal
2)
∧
provides(?
agent
2
,
?
mitigateRisk
)
∧
6 : hasNegImpact(?
risk,
?
g
1)
→
7:
sqwrl
:
select
(?
team,
?
goal,
?
agent
1
,
?
g
1
,
?
agent
2
,
?
g
2
,
?
risk,
?
mitigateRisk
)
∧
hasNegContribution(?
mitigateRisk,
?
risk
)
∧
8:
sqwrl
:
columnNames
(”
team
”
,
”
goal
”
,
”
agent
1”
,
”
g
1”
,
”
agent
2”
,
”
g
2”
,
”
risk
”
,
”
mitigateRisk
”)
Fig. 4.
SQWRL representation of DP 2.1
aim(?
a,
?
goal
)
←
requests(?
a,
?
goal
)
aim(?
a
2
,
?
goal
)
←
requests(?
a
1
,
?
goal
)
∧
hasDelegation(?
a
1
,
?
d
)
∧
hasDelegatee(?
d,
?
a
2)
∧
hasDelegatum(?
d,
?
goal
)
aim(?
a
2
,
?
goal
)
←
aim(?
a
1
,
?
goal
)
∧
hasDelegation(?
a
1
,
?
d
)
∧
hasDelegatee(?
d,
?
a
2)
∧
hasDelegatum(?
d,
?
goal
)
Fig. 5.
Relaxing “request” on SI* in SQWRL
of “request” is more relaxed - direct request (i.e., the actor “requests” fulfillment of a
goal) or indirect request (i.e., another actor delegates the execution of a goal to him/her,
to the actor). Accordingly, we decided to extent the DL TBox and revise the pattern
formalization using those new concepts/roles. However, this extension can only be done
when we use SQWRL and
not
SPARQL. In Fig. 5, we illustrate an example of the
extension of the “request” relation in SI*, namely “aim”. To be closer with the DP2.1's
description one needs to replace line 3-4 of Fig. 4 with the following:
aim(?
agent
1
,
?
g
1)
∧
aim(?
agent
2
,
?
g
2)
4.4
Representing the Problem in the ABox
Finally, designers need to represent the problem-at-hand in terms of instances of con-
cepts and roles in the ABox.
Concept versus Individual.
Individuals have a unique identity, and their description
can be modified by adding more assertions in the ABox. Conversely, the definition of
concepts cannot be modified [16].
The first alternative (i.e., subclasses of the DL TBox) allows us to reason whether a
pattern appears anywhere in the problem, but it cannot provide the mapping between
construct in the pattern's solution and the problem-at-hand. The second alternative (i.e.,
individuals in the DL ABox) on the other hand can provide such mappings, but it will
not allow us to reason in a situation where the problem contains both abstract and con-
crete entities in the real world, because both entities will be encoded as individuals and
the reasoner will treat them equally. Since we deal with the problem at the design level
where mostly models capture the class level instead of the object one, we have chosen
the second alternative (i.e., as series of individuals) as the most suitable to our needs.
