To assess the coverage area we selected a set of visualization techniques
described in the scientific literature or widely used in the 3D GIS domain, and
created for each of them a representation using the vocabulary defined in the
ontology. The experience showed that the defined vocabulary was sufficient to
describe those techniques. This is obviously a partial evaluation since there are
still other techniques. Adequacy assessing should check the adequacy for the
intended application. More precisely it should check that an application searching
visualization techniques and detecting inconsistencies is more efficient when using
this ontology rather than when using a database, for example.
We have limited the evaluation to the effectiveness of the ontology, i.e. to ver-
ify that an application that uses this ontology can really (1) Find 3D visualization
techniques corresponding to complex research criteria (2) Detect inconsistencies
between techniques. For this, we defined several queries considered as representa-
tive for a designer of 3D urban models, as well as cases of incompatibility to detect.
We present below an example of query and an example of inconsistency detection.
5.1 Querying the Knowledge Base
The main query task consists in finding techniques that are suitable for a particu-
lar dataset and context of use. This amounts to describe the characteristics of the
desired technique in the form of a class specification and to let the logical reasoner
infer which techniques belong to this class.
The following expression is intended to find techniques to represent integer val-
ues that are located at specific points in a grid.
and dataDescription some (DataDescription
and datatype some IntegerType
and coverage some Point
and containedIn some Grid)
Since RealNumberType has been defined as a subclass of IntegerType , this query
will also return techniques that can represent real numbers. This is what we want
since a technique that can display real numbers will certainly work for integer
numbers. And this is precisely what an OWL reasoner will do.
Similarly, this query will also return techniques that have more specific cover-
age or containedIn properties such as, for instance, a technique to display numbers
that are located at points in an orthogonal grid (a special type of grid).
The following example query finds techniques for the estimation of air pol-
lutant concentration with a navigation context above the scene or in the scene (at
and (usage some ((task some Estimate) and (usage some (Context