Geoscience Reference
In-Depth Information
represented in this class corresponds to forests of area greater than 5
hectares in the real world.
The remainder of the paper is organized as follows: first, a presentation of
our approach and an introduction to the elements necessary for the
understanding of the system is done. Then, the system architecture is
detailed, and finally the implementation of a prototype and some results
obtained with
IGN
data are showed.
2- Approach and methods
2.1 Preliminary notions
For the sake of clarity, let us first describe the following core elements of
our system: the geographic database specifications, the topographic
domain ontology and the specification ontologies.
2.1.1 Geographic databases specifications
Like any database, geographic databases are described by their schema.
Classes are named by common geographic words, which usually refer to
geographic concepts. Their instances, geographic features, are described
by attributes and a geometrical representation (usually point, line or
polygon). However, each geographic database represents the specific point
of view of its producer about the geographic real world (Fonseca et al.
2003). As an example, if a class is named '
Road
', it may actually
designate only hard-surface roads, or alternatively include loose-surface
roads, such as trails, footpaths or forest rides; moreover, it may or not
designate only main roads with at least two lanes. Actually, a geographic
database is associated with a certain level of detail. Therefore, only
relevant geographic features are captured in this database. Besides, in
vector databases, the geometric representation of a given geographic
feature may vary: a road may be represented by a line drawn along its axis,
or by a polygon representing the surface it covers.
Since data capture for a given database is often done by several persons,
special attention has to be paid to the homogeneity of data meaning within
