Geoscience Reference
In-Depth Information
Here
wn:pedestrian
and
wn:count(n1)
are classes imported from the WordNet
ontology. Since the word
count
has several senses, the
(n1)
suffix indicates that we
refer to the first meaning of count as a noun (the total number counted).
4.3 Incompatibility Rules
The term incompatibility defines the impossibility to use several visualization
techniques together. For instance, a texture rendering technique cannot a priori be
used with another technique with the same type of rendering because there is a
risk of occlusion and thus of confusion when interpreting the information in case
they are attached to the same city object. However a texture rendering technique
can be potentially used with a technique having a different type of rendering (dis-
play panel for example).
The detection of incompatibilities between techniques is difficult or even
impossible to express in the OWL language. However, they can be expressed as
inference rules in the SWRL rule language, SWRL being a combination of OWL
and RuleML expressing rules on ontological entities (Horrocks et al.
2004
). Using
such rules, an OWL reasoner is able to detect possible incompatibilities between
visualization techniques.
The rules that we have defined are based on real test cases of several 3D visu-
alization techniques used inside the same 3D city model. For example, we have
defined a rule expressing that two techniques t1 and t2 are potentially incompat-
ible if their output location is on objects of the same type and if their output shapes
are the same (e.g. they both use boxes on buildings). This rule can be written as
follows in (simplified) SWRL:
outputLocation(t1, ol1)
and
outputLocation(t2, ol2)
and
objectType(ol1, ot1)
and
objectType(ol2, ot2)
and
ot1
=
ot2
and
objectRelation(o1, r1)
and
objectRelation(o2, r2)
and
r1
=
r2
and
outputShape(t1, s1)
and
outputShape(t2, s2)
and
s1
=
s2
→
incompatiblityInOutputLocationAndShape(t1, t2)
More sophisticated incompatibility rules could be defined, in particular to take into
account criteria related to spatial or topological relations.
5 Evaluation of the Ontology
The type of ontology that we developed can be evaluated according two ways: the
coverage area (data-driven evaluation) and the adequacy for the intended applica-
tion (application-based evaluation) (Brank et al.
2005
).
