- From: Boris Villazón Terrazas <bvillazon@fi.upm.es>
- Date: Tue, 18 Jan 2011 10:51:29 +0100
- To: Giovanni Tummarello <giovanni.tummarello@deri.org>
- CC: public-lod <public-lod@w3.org>
Hi Giovanni I'll try to do my best to explain the potential benefits of this exposure of geometry with an example. For example, let's suppose that you are going for holidays to visit the Mora Municipality in Spain. Since you practice water-related sports, you may want to know resources for this within a specific distance (bounding box) of this municipality (e.g., beaches, lakes, rivers or something similar). In the geospatial context, the transformation and publication of the OpenStreetMap [1], Ordnance Survey [2], and GeoLinkedData [3] data according to the Linked Data principles have added a new dimension to the Web of Data. In this way spatial data can be retrieved and interlinked. For our example, we query the GeoLinkedData [3] for that purpose. GeoLinkedData [3] has geospatial resources like rivers, lakes, reservoirs, etc. Most of these resources have associated geometry objects. These geometry objects consist basically of points, and linestrings, for example, the course of a river. SPARQLing for them and putting them on a map isn't too hard. So, you can express this query in SPARQL. Next, the (inefficient) query without geospatial support PREFIX geo: <http://www.w3.org/2003/01/geo/wgs84_pos#> SELECT ?subject ?label ?latitude2 ?longitude2 WHERE { <http://geo.linkeddata.es/resource/Municipio/Mora> <http://www.w3.org/2003/01/geo/wgs84_pos#geometry> ?g. ?g geo:lat ?latitude. ?g geo:long ?longitude. ?subject <http://www.w3.org/2003/01/geo/wgs84_pos#geometry> ?g2. ?g2 geo:lat ?latitude2. ?g2 geo:long ?longitude2. ?subject rdfs:label ?label. FILTER(xsd:double(?latitude2) - xsd:double(?latitude) <= 0.1 && xsd:double(?latitude) - xsd:double(?latitude2) <= 0.1 && xsd:double(?longitude2) - xsd:double(?longitude) <= 0.1 && xsd:double(?longitude) - xsd:double(?longitude2) <= 0.1 && lang(?label) = "es"). } limit 20 Next, a similar query with RDF-based geospatial support PREFIX geo: <http://www.w3.org/2003/01/geo/wgs84_pos#> SELECT ?subject ?latitude2 ?longitude2 (bif:st_distance(bif:st_point(?longitude,?latitude), bif:st_point(?longitude2,?latitude2)) ) as ?distancia ?latitude ?longitude WHERE { <http://geo.linkeddata.es/resource/Municipio/Mora> geo:geometry ?g . ?g geo:lat ?latitude. ?g geo:long ?longitude. ?subject geo:geometry ?g2. ?g2 geo:lat ?latitude2. ?g2 geo:long ?longitude2. } limit 20 As you may know there are some SPARQL extensions for geometry and geographical data, and people are using that. You can check the Virtuoso mailing list to see how many people are asking/using the Virtuoso geospatial features. And there are lots of additional queries that you can do, for example: distance between two resources, intersection with respect a point, etc. Of course, this does not necessarily have to substitute (either partially or completely) the native support that many geographical databases already provide, and for which they are very well designed and tested in production environments. The situation is that sometimes it may prove to be interesting to have (some of) this data in RDF as well, so that inclusion in SPARQL queries is easier. It may be considered similar to why we make transformations of (partial or complete) relational databases into RDF. I hope I explained well. Boris [1] http://www.openstreetmap.org/ [2] http://www.ordnancesurvey.co.uk/ [3] http://geo.linkeddata.es/ On 17/01/2011 14:07, Giovanni Tummarello wrote: > Boris would you be able to provide a bit of explanation on why would > you want o do that e.g. what evidence are there (nice use cases) were > an rdf export of low level features in the map is of use > thanks! > Gio > > On Mon, Jan 17, 2011 at 2:34 AM, Boris Villazón Terrazas > <bvillazon@fi.upm.es> wrote: >> Victor, Miguel Angel and Boris >> >>
Received on Tuesday, 18 January 2011 09:51:58 UTC