[Requests] Comments on candidate GeoSPARQL standard (OGC 11-052r3)

Simon.Cox at csiro.au Simon.Cox at csiro.au
Fri Aug 5 04:24:21 EDT 2011


PART A

1. Evaluator:
Simon Cox
Simon.cox at csiro.au

2. Submission:
OGC GeoSPARQL - A Geographic Query Language for RDF Data (11-052r3)

PART B

I) RDF Namespace
1. Requirement: Editorial/Technical
2. Implementation Specification Section number: 6.2, 8.2, 8.3, 8.4
3. Criticality: Major
4. Comments/justifications for changes:
The specification proposes a number of ontology classes and properties. 
The identifiers for these are put in the RDF namespace "http://www.opengis.net/ont/OGC-GeoSPARQL/1.0". 
This is inconsistent with the existing OGC URI Policy (there is no 'ont' name-type) and also with external best-practice, such as described in the UK Public Sector document http://www.cabinetoffice.gov.uk/resource-library/designing-uri-sets-uk-public-sector 
These policies are consistent in recommending that OGC Ontology URIs (i.e. URIs for RDF descriptions) have the structure http://www.opengis.net/def/{name}. 
Given the practical need to partition the namespace after /def/ to help with namespace management, but also a desire to simplify the current OGC URI namespace to better fit in with semantic-web practice in the broader external community, perhaps a simple split like this would suffice:

http://www.opengis.net/def/class/{name}
http://www.opengis.net/def/property/{name} 

The rdfs:isDefinedBy property can be used to tie each definition back to the document in which it is defined (e.g. GeoSPARQL 1.0), so it may not need to be reflected in the URI. 

NOTE: I acknowledge that I was party to this design, but further investigation of external community practice leads me to suggest this change. 


II) Use of OWL
1. Requirement: Technical
2. Implementation Specification Section number: 8.3, 8.4
3. Criticality: Major
4. Comments/justifications for changes:
The semantic elements are defined using RDFS. Some use of OWL would improve the specification, in particular the standard properties could useful be asserted to be of type owl:AnnotationProperty, owl:DatatypeProperty and owl:ObjectProperty as appropriate. 


III) Relationship with OGC Abstract Specification (ISO/TC 211 Harmonized Model)
1. Requirement: Technical
2. Implementation Specification Section number: 6.2, 8.2
3. Criticality: Major
4. Comments/justifications for changes:
It would be expected that the classes SpatialObject, Feature and Geometry have a relationship with the UML classes GFI_Feature and GM_Object defined in ISO 19109/19156 and ISO 19107. This should be explained. 



Simon Cox | Research Scientist
CSIRO Earth Science and Resource Engineering 
ARRC, PO Box 1130, Bentley WA 6102, Australia 
Tel +61 8 6436 8639 | Mob +61 403 302 672
simon.cox at csiro.au | http://www.csiro.au/people/Simon.Cox



More information about the Requests mailing list