Alignment to BFO is seemingly not satisfiable

[avsculley]

Greetings,

I am currently working on revising your alignment between BFO and GeoSPARQL. I am hoping that we can work together to some degree so that I can get some clarity on a few things.

I will begin with this quote from the alignment, which seems to make geo:Geometry unsatisfiable:

geo:Geometry rdfs:subClassOf obo:BFO_0000006 .
geo:Geometry rdfs:subClassOf obo:IAO_0000030 .

It seems unsatisfiable because obo:BFO_0000006 is a subclass of independent continuant and obo:IAO_0000030 is a subclass of generically dependent continuant. Further, Independent continuant and generically dependent continuant are disjoint classes in BFO. Thus, geo:Geometry cannot have an instance.

It is possible that it just seems unsatisfiable if, for example, the intention of the alignment was to use just one or the other mapping. Was the intention behind the alignment to use the whole thing with BFO?

[paulc-dstl]

This is an interesting observation, and one which I think relates to the difference between a spatial region and representations of a spatial region. A representation of a spatial region is data, an information content entity if you like, describing a region of space(time). Geometries in GeoSPARQL (and serialisations thereof eg WKT, geojson, etc) are therefore representations (sets of coordinates in some coordinate space describing some region), but given the definition of "spatial region" in BFO includes "a coherent set of direction positions in space", it appears that BFO's spatial region can be both the region in space and the set of coordinates describing it. Hence the mapping as shown. But a disjoint relationship in parent BFO classes would indeed cause problems, as you note.

Do you have some examples of coordinate geometries currently used in a BFO conformant way which might help enlighten this discussion?

Many thanks!
P.

NB I didn't do the GeoSPARQL-BFO mapping, but am currently working on some other BFO related activities for which this Issue is informative, hence my interest.

[avsculley]

Bfo:spatial region is just the region in space, not the information that represents it. We use either IAO or CCO Information Content Entity to model representations of regions in space. From my reading of GeoSPARQL documentation, it seemed to me that geo:Geometry could be either the region in space or the information representing it. What is the reason to think that geo:Geometry is just the representation of a region in space, and not also the region in space itself? (Perhaps a naïve question - my background is in BFO ontologies, not GIS.)

I don't have any examples for you, unfortunately. But if I find any I will let you know.