Database Reference
In-Depth Information
11 <geo:lat>51.730031</geo:lat>
12 </rdf:Description>
13 </rdf:RDF>
The GeoOnion RDF/XML vocabulary,
15
which takes the namespace “
go:
,” was
an early attempt, as yet incomplete, to provide a number of properties that relate
spatial things together based on their distance in meters. It suffers from a lack of GI
input, as the initial idea was to specify nearness in terms of concentric circles around
the point of interest. This is a woefully inadequate way to deal with the richness of
the nearness relationship. It is in exactly this kind of initiative that the GI specialist's
experience of the contextual dependence of geographical nearness could come into
play in modeling knowledge more accurately.
In addition, there are several OWL ontologies that deal with spatial information,
such as the geoFeatures,
16
geoCoordinateSystems,
17
and geoRelations
18
ontologies.
5.8 RDF FOR THE THINKING GEOGRAPHER
So, from the geographer's point of view, what are the most important differences
between RDF and traditional methods of representing data in the geographical
field? As we have seen, the RDF data model is a graph, which makes it far more
amenable to the addition of new information than the more familiar tabular form
used to represent data in a GIS sitting on top of a relational database. While adding a
new property to a geographical feature (that is, adding a new column to a relational
database table) means that every geographical feature must provide a value for that
field, or include a null, thus increasing the size of the table significantly, this is not
the case in RDF, where only new triples need be added to the graph, without the need
to pad a table.
Identity is another topic that the geographer must treat carefully. A real-world
object may well be given an identifier in a GIS, and there have been efforts made to
create a global system of identifiers. However, RDF has a ready-made, and superior,
system for minting identifiers through URIs. In many cases, an organization can use
its own internal system for identifier creation, plus their Web domain name, to create
public URIs for their data. The process of discerning whether two things are the
same is an outstanding problem both for the Semantic Web and for a GIS specialist.
While in GIS spatial collocation is often used as the primary factor in determining
an equivalence relationship, this is far less likely to be a sufficient indicator in the
semantic world. For example, a building in a dataset mapped 100 years ago with
exactly the same spatial footprint as the building today might well be identified as
the same building by a GIS even though its usage had changed. However, RDF might
have two URIs, one for a Prison, which the building was used as last century, and
one for a Hotel, which is the building's current use. RDF could also have a third URI
that represents the building irrespective of its use. RDF places far more emphasis on
context
, whereas a GIS is mainly concerned with spatial information.
As we explain in Chapter 9, using the
owl:sameAs
relationship between two
URIs is quite a strong assertion to make as
all
the statements about one URI are then
known to be true about the equivalent URI. However, room prices at the hotel are
