Database Reference
In-Depth Information
Virtuoso enables the representation of features in a two-dimensional space, by
realizing a new geometry data type (virtrdf:Geometry) and a corresponding
R-Tree index, and in this way allows the use of spatial operators (actuall SQL
MM functions) in SQL and SPARQL queries
20
. This feature was proprietary in
the version 6 of Virtuoso that was exploited by our previous work but it can now
be fully exploited in our new version of the LD Management system proposed
in this article. As it will be seen in the next section, through this new feature,
the user has now the capability to perform geospatial queries in two different
ways, where the inline Virtuoso way has a better performance but lacks particu-
lar geospatial functionality. In this way, the user has the ability to choose one of
these two ways depending on the desired geospatial functionality to be exploited
in the respective queries and his/her query performance requirements.
Fully supporting a set of geospatial functions is not common in the corre-
sponding systems. The available standards like GeoSPARQL are rather young
and the respective systems offer limited support, as discussed also in the previ-
ous sections. Virtuoso on the other hand provides a very good infrastructure to
extend and build our own functionalities.
4 LD Management Service Extensions
While the initial functionality of LMS focused mainly on providing basic LD
management functionality, it became apparent that there was a need for also
appropriately handling the special LD type of geospatial data. Apart from this,
additional importing requirements were set based on the feedback from particu-
lar data providers in the InGeoCloudS project, while some performance problems
were identified due to the use of Strings as the main representations of the LMS
methods' input and output (I/O). Moreover, due to the fact that it is imposed
that the INSPIRE directives should be embraced by data providers until 2020, it
was also decided to realize INSPIRE exporting functionality for the LD stored.
To this end, LMS was extended according to the following aspects: (a) addi-
tional methods were introduced which enable the management of geospatial LD,
(b) an additional import method was introduced, (c) methods for exporting LD
in INSPIRE format were implemented and (d) the handling of I/O now occurs
via java streams. The additional methods realized are the following:
-
geoldquery:
This method is similar to
ldquery
but allows the issuing of
GeoSPARQL queries instead of SPARQL ones. The realization of this method
relied on the exploitation of the USeekM framework
21
.
-
geoldtransform:
This method can be used to transform (Geo)SPARQL results
of any format into a feature collection representation format, such as GML
22
and KML
23
. As some of these formats enable a direct representation of the
Search WWH ::
Custom Search