CRMgeo for Getty TGN representation

Posted by Vladimir 22/4/2014 

Hi everyone! 
I've been digging through CRMgeo as one potential representation for Getty TGN. 

I'm working with these docs: 
– http://www.cidoc-crm.org/docs/Technical%20Report435-CRMgeo.pdf 
– http://www.ics.forth.gr/isl/CRMext/CRMgeo/docs/CRMgeo.pptx 
– http://www.ics.forth.gr/isl/CRMext/CRMgeo_v1.0.owl 

I attach a cleaned Turtle that's easier to read (for people who rarely use Protègè). In particular, I've corrected the prefixes. 
I also attach the two key diagrams from the presentation (excellent!) 
In particular, look at CRMgeo-GeoSPARQL-Integration.jpg andPhenomenal_vs_Declarative_Spacetime_Volumes.jpg 
1. It would be good to define domain & range for all properties (e.g. Q7i doesn't have them). owl:inverseOf does not automatically make any assertion about range & domain from those of the forward property. 

2. The spec quotes GML time expressions, but geo:Geometry is grafted under crmgeo:SP5_Geometric_Place_Expression, with no relation to Time Expression. 
I don't have experience with GML time expressions, TGN doesn't need them, and I plan to use WKT only... 
But is this intentional? 

3. If 2 is intentional, I would expect crmgeo:Q10i_place_is_defined_by to be related to geo:hasGeometry ? 
But there's no path between them, without going through entities with more info (Period or Physical Thing) 

4. crmgeo:SP5_Geometric_Place_Expression is subclass of crm:E47_Spatial_Coordinates, which has direct link to Place (crm:P87i_identifies). 
Is it ok/permissible to use this link, should it be considered a shortcut of Q10+something? 

5. The deeper question is that Place (just like Time-Span) is not considered geo:Feature. 
Looking at the other slide, geo:Feature is equated to "real-world phenomenon", whereas Place is not considered such, in line with Issue 201 "there is no more substance to a place than the area". 
This is not intuitive to me, since many other ontologies consider Places to be Features (e.g. GeoNames, LOCN) 
Do I understand this right? 

6. Could you please comment on the representation below: 

tgn:3000034-place a crm:E53_Place, crmgeo:SP6_Declarative_Place; 
crm:P87_is_identified_by tgn_term:1234-en; 
crmgeo:Q10i_place_is_defined_by tgn:3000034-point, tgn:3000034-box; 
crm:P3_has_note "Refers to the area around the Great Lakes, which..."@en; 
crm:P2_has_type aat:1234; # "general region" 
crm:P89_falls_within tgn:1000001-place. # "North and Central America" 

tgn_term:1234-en a xl:Label, crm:E48_Place_Name; 
xl:literalForm "Great Lakes Region"@en; 
crm:P3_has_note "Great Lakes Region"@en; 
# sources, contributors, etc. 

tgn:3000034-point a crmgeo:SP5_Geometric_Place_Expression, sf:Point; 
wgs:lat "45.0000"; wgs:long "-85.0000"; wgs:alt "183.1840"; 
geo:asWKT "Point(-85.0000 45.0000)"^^geo:WKTLiteral. 

tgn:3000034-box a crmgeo:SP5_Geometric_Place_Expression, sf:Polygon; 
geo:asWKT "Polygon((-92.0160 43.1560, -92.0160 48.8120, -82.4910 48.8120, -82.4910 43.1560, -92.0160 43.1560))"^^geo:WKTLiteral. 

7. The previous proposal on which CRMgeo builds upon is at: 
http://crmeh.files.wordpress.com/2012/01/coordinates2crm_v2.pdf 
http://geosparqlworkshopyork080212.wordpress.com/geo-sparql-workshop-yo…;

It refers to ontology file GeoSPARQL_4_CRMv2.owl 

Could one of the authors (Gerald Hiebel, Øyvind Eide, Mark Fichtner, Klaus Hanke, Georg Hohmann, Siegfried Krause) please send it? 

Posted by Gerald Hiebel 2/05/2014 

Hi Vladimir, 
Thanks for the turtle code. 
I put my comments in your text, Very interesting issues, 
Best, 
Gerald 

[...]

1. It would be good to define domain & range for all properties (e.g. Q7i doesn't have them). owl:inverseOf does not automatically make any assertion about range & domain from those of the forward property.

Thanks, good idea 

2. The spec quotes GML time expressions, but geo:Geometry is grafted under crmgeo:SP5_Geometric_Place_Expression, with no relation to Time Expression. I don't have experience with GML time expressions, TGN doesn't need them, and I plan to use WKT only... But is this intentional?

It is not intentional. We only introduced Time Expression and Spacetime volume Expression after making the alignment to GeoSPARQL. I believe they should be subclasses of Geometry too. If we want to be more precise we have to go into the GML tree and assign them to the subclasses related to time for the time expression. 
 

3. If 2 is intentional, I would expect crmgeo:Q10i_place_is_defined_by to be related to geo:hasGeometry ? 
But there's no path between them, without going through entities with more info (Period or Physical Thing)

Have to take a look after solving 2. 

4. crmgeo:SP5_Geometric_Place_Expression is subclass of crm:E47_Spatial_Coordinates, which has direct link to Place (crm:P87i_identifies). 
Is it ok/permissible to use this link, should it be considered a shortcut of Q10+something?

Depending if it goes to a SP2 Phenomenal Place(Shortcut for crm:E47_Spatial_Coordinates crmgeo:Q10 defines: crmgeo:SP6 Declarative Place crmgeo:Q11 approximates crmgeo:SP2 Phenomenal Place) or SP6 Declarative Place where it should be equivalent to crmgeo:Q10 defines. 
 

5. The deeper question is that Place (just like Time-Span) is not considered geo:Feature. 
Looking at the other slide, geo:Feature is equated to "real-world phenomenon", whereas Place is not considered such, in line with Issue 201 "there is no more substance to a place than the area". 
This is not intuitive to me, since many other ontologies consider Places to be Features (e.g. GeoNames, LOCN) Do I understand this right?

This is truly a deeper question, which is related to the problem that in the definition of feature in the end a feature can be anything and this does not help in an ontology. I believe the ontological separation of Place and the thing occupying the Place or defining the place makes sense for a lot of questions but in specific cases it is also usefull to make a double instantiation as proposed in the latest version of CRMsci (http://www.ics.forth.gr/isl/index_main.php?l=e&c=663) where S20 Physical feature is a subclass of E26 and E53. 

But this is probably better discussed than in a mail and you are right it's a quite fundamental question. 

For the definition of feature see also: 

OGC. 2009, The OpenGIS Abstract Specification: Topic 5: Features. http://www.opengeospatial.org/standards/as, http://portal.opengeospatial.org/files/?artifact_id=29536 

6. Could you please comment on the representation below: 
tgn:3000034-place a crm:E53_Place, crmgeo:SP6_Declarative_Place; 
 

SP6 Declarative Place sounds right to me. The SP6 Declarative Place is defined by something like a "Natural Language Place Expression": ("area around the Great Lakes"). We considered in the start to include these expressions in CRMgeo, but would have needed more research on that. Øyvind knows more about these things, but they are not part of CRMgeo (yet). In the end what they do is define SP6 Declarative Places or point to SP2 Phenomenal Places. 
 

crm:P87_is_identified_by tgn_term:1234-en; 
crmgeo:Q10i_place_is_defined_by tgn:3000034-point, tgn:3000034-box;

The coordinates tgn:3000034-point, and tgn:3000034-box; do not define the SP6 Declarative Place but approximate the place that is defined (crmgeo:Q10i_place_is_defined_by or a new property) by a "Natural Language Place Expression". So in this case we have 3 Declarative Places. The one identified by tgn:3000034 and meaning the area around the Great Lakes and two SP6 Declarative Places tgn:3000034-point, tgn:3000034-box that approximate tgn:3000034. 

crm:P3_has_note "Refers to the area around the Great Lakes, which..."@en; 
crm:P2_has_type aat:1234; # "general region" 
crm:P89_falls_within tgn:1000001-place. # "North and Central America"

ok

tgn_term:1234-en a xl:Label, crm:E48_Place_Name; 
xl:literalForm "Great Lakes Region"@en; 
crm:P3_has_note "Great Lakes Region"@en; 
# sources, contributors, etc.

ok

tgn:3000034-point a crmgeo:SP5_Geometric_Place_Expression, sf:Point; 
wgs:lat "45.0000"; wgs:long "-85.0000"; wgs:alt "183.1840"; 
geo:asWKT "Point(-85.0000 45.0000)"^^geo:WKTLiteral.

ok

tgn:3000034-box a crmgeo:SP5_Geometric_Place_Expression, sf:Polygon; 
geo:asWKT "Polygon((-92.0160 43.1560, -92.0160 48.8120, -82.4910 48.8120, -82.4910 43.1560, -92.0160 43.1560))"^^geo:WKTLiteral.

ok 
An interesting issue that I see: the TGN record has also: 
Related geographic places: 
related to .... Great Lakes .......... (lakes) 

.................. (World, North and Central America) [7029370] 

And this relates now to a SP2 Phenomenal Place, in fact Great Lakes(tgn:7029370) crm:P89_falls_within Great Lakes Region (tgn:3000034) 

That's my fast guess, comments from Øyvind and Martin would be needed. 

[…]

posted by Ritchard Light  on 10/2/2015

FYI.

Richard

-------- Forwarded Message --------
Subject:  Linked Geospatial Data use cases
Resent-Date:  Mon, 9 Feb 2015 22:10:59 +0000
Resent-From:  public-lod@w3.org
Date:  Mon, 9 Feb 2015 22:11:53 +0000
From:  Phil Archer <phila@w3.org>
To:  LOD List <public-lod@w3.org>

Dear all,

As I hope you'll have seen, we recently launched a new Spatial Data on
the Web Working Group [1]. We (W3C) are collaborating closely with our
sister standards body, the Open Geospatial Consortium, so that the
standards we produce will be jointly published by both OGC and W3C. In a
nutshell, the idea is to make GIS and LOD work together more easily.

The WG is collecting use cases very rapidly [2] - this WG is being very
productive very quickly. But we need your help. The conversation and the
use cases so far are much more heavily weighted towards the world of OGC
and GIS than Linked Data.

If you work with spatial data (including non-geo uses such as in life
sciences) and do so in a Linked Data framework, if you use LD to
integrate disparate data sources and link them to a location etc. then
please let us have your use cases.

Ideally of course you join the WG, but if you can't so that, we'd like
to hear your use cases via the public-sdw-comments@w3.org. Obviously use
cases that cover areas not already covered are the ones we're most
hoping for.

Decisions are made by those who turn up... and we really want you to
turn up!

Thanks

Phil.

[1] http://www.w3.org/2015/spatial
[2] https://www.w3.org/2015/spatial/wiki/Working_Use_Cases

--

Phil Archer
W3C Data Activity Lead
http://www.w3.org/2013/data/

http://philarcher.org
+44 (0)7887 767755
@philarcher1