Adding new property hasFirstMember. Resolves #549.

[dylan-sa]

No description provided.

[rjyounes]

This looks nearly perfect. We didn't mention this at the meeting, but it occurs to me you could also add a restriction on OrderedCollection to the effect that it is either empty or has a first member. (Note that due to OWA the first member doesn't have to be stated.)

A little nitpicky, but most commonly I see the OWA spelled "Open World Assumption" rather than the hyphenated lowercase version, though I know this is also used.

[dylan-sa]

For OrderedCollection, maybe something like the following would work? If so, should I go ahead and include in this PR?

gist:OrderedCollection
	a owl:Class ;
	owl:equivalentClass [
		a owl:Class ;
		owl:intersectionOf (
			gist:Collection
			[
				a owl:Restriction ;
				owl:onProperty gist:hasMember ;
				owl:allValuesFrom gist:OrderedMember ;
			]
			[
				owl:unionOf (
					[
						owl:complementOf [
							a owl:Restriction ;
							owl:onProperty gist:hasMember ;
							owl:someValuesFrom gist:OrderedMember ;
						] ;
					]
					[
						a owl:Restriction ;
						owl:onProperty gist:hasFirstMember ;
						owl:someValuesFrom gist:OrderedMember ;
					]
				) ;
			]
		) ;
	] ;

[uscholdm]

I'm not sure I follow; it looks like it is literally saying the following:

skos:definition "A collection that only has ordered members and either has a first member that is an ordered member or has no members that are ordered members."^^xsd:string ; 

Is this your intention? A slightly simpler alternative to complement would be to use max cardinality of 0

[rjyounes]

The complement is not correct: you need to use owl:Thing instead of gist:OrderedMember to say that the set has no members. Otherwise the complement includes things which have objects other than OrderedMembers as members.

[rjyounes]

See comments.

[dylan-sa]

@rjyounes Ah, that's right! I will go ahead and work the correct version into the PR.

[rjyounes]

Actually, my previous comment may be incorrect given the first restriction. However, why not reduce the nesting - and impro ve the comprehensibility - as follows?

	[

			a owl:Restriction ;
			owl:onProperty gist:hasMember ;
			owl:cardinality "0"^^xsd:nonNegativeInteger ;
	]

And since the range of hasFirstMember is OrderedMember, you can also use a cardinality restriction:

			[
				a owl:Restriction ;
				owl:onProperty gist:hasFirstMember ;
				owl:cardinality "1"^^xsd:nonNegativeInteger ;
			]

Or, if you prefer, since hasFirstMember is a functional property and thus the cardinality is inferred:

			[
				a owl:Restriction ;
				owl:onProperty gist:hasFirstMember ;
				owl:someValuesFrom owl:Thing
			]

Which makes me realize that hasFirstMember should be declared a functional property.

[uscholdm]

Which makes me realize that hasFirstMember should be declared a functional property.

good point.

[dylan-sa]

I have no objection to going the functional route, but in #549 it looks like we wanted to allow for "first-place ties," so to speak.

[rjyounes]

Good point, you're right, we do want first place ties, so not functional.

[rjyounes]

But it now occurs to me that it's an inverse functional property, since each OrderedMember belongs to a unique OrderedCollection and points to a unique object.

[uscholdm]

I think that's right, given an ordered member is just the proxy for the real thing.

[dylan-sa]

OK, I think the latest two commits get it right. hasFirstMember is now inverse functional, and I've tweaked the restriction for OrderedCollection in line with discussion above. (Namely, I went with the 'cardinality 0' construction over the clunky 'complement' construction.) If this looks good, I will rebase to catch up with the latest changes to develop.

[rjyounes]

Everything you've done is perfect. Can you add one sentence to the scope note along the lines of "Since ordered collections may not be strictly ordered, there can be more than one first member."

After that it's good to go.