Information Technology Reference
In-Depth Information
prefix tropos :
<
http ://www.owl
−
ontologies .com/ Tropos Ontology .owl#
>
SELECT
?team ?goal ?agent1 ?agent2 ?subgoal1 ?subgoal2
WHERE
{
?team tropos : request ?goal .
?goal tropos : isAndDecompositionOf ?g1.
?goal tropos : isAndDecompositionOf ?g2.
FILTER (? subgoal1 != ?subgoal2 ).
?agent1 tropos : isPartOf ?team.
?agent2 tropos : isPartOf ?team.
FILTER (? agent1 != ? agent2 ) .
?team tropos : hasDelegation ?d1.
?d1 tropos : hasDelegatee ?agent1 .
?d1 tropos : hasDelegatum ?g1.
?team tropos : hasDelegation ?d2.
?d2 tropos : hasDelegatee ?agent2 .
?d2 tropos : hasDelegatum ?g2.
?agent2 tropos : provides ?mitigateRisk .
?g1 tropos : hasNegDContribution ?g2.
?task tropos :hasNegContribution ?risk .
?risk tropos :hasNegImpact ?g1.
OPTIONAL
{
?agent2 tropos : requests ?mitigateRisk.
}
}
Fig. 3.
SPARQL representation of DP 2.1
Turtle data format
2
is used to represent triple patterns. The query attempts to match the
triples on the graph pattern against the model [14]. SPARQL just queries the model and
does not support inference [12], nor does it modify the RDF dataset. However, some
frameworks (e.g., JENA) and rule engines [15], have the capacity to perform inference
and update the dataset by performing OWL reasoning.
Alternatively, a more expressive query language that is founded on DL semantics
and supports comprehensive querying of OWL is SQWRL [13]. SQWRL is a SWRL-
based query language [7]. SQWRL provides SQL-like operations to retrieve knowledge
from an OWL ontology. Similarly to SPARQL, in SQWRL we try to capture all con-
cepts and relationships present in a pattern. Since SQWRL understands the semantics of
OWL and SWRL rules, it understands not only the explicit, but also the inferred knowl-
edge. For example, the DP2.1 of SERENITY pattern (Fig. 2 described in Table 1), can
be translated into a SPARQL Query (Fig. 3). Each node of the pattern context is a vari-
able in the query and each edge is an RDF triplet. For a SQWRL Query, the DP2.1
translational is shown in Fig. 4.
4.3
Enriching DL T-Box with Implicit Knowledge
Often details of the patterns are described in natural language, due to the expressivity
limitation of the pattern language. This was certainly the case with our case study.
Back to our example in DP2.1, in SI* the notion of “request” means that an actor
intends to achieve a particular goal. However, based on DP2.1's description the intent
2
Turtle:
http://www.w3.org/TeamSubmission/turtle/
