531 extend symbol predicates

[uscholdm]

Fixes #531 .

• Deprecated predicates: gist:unitSymbol, gist:unitSymbolHtml and gist:unitSymbolUnicode.
• Added predicates: gist:symbol, gist:symbolHtml and gist:symbolUnicode.
• Added symbol triples for existing units, anticipating the removal of gist:UnitSymbol.
• Added nonConformingLabel triple for symbolUnicode in gistValidationAnnotations.ttl

[rjyounes]

I've suggested what I think are more precise definitions; they could still use more work. We should also settle on whether to put letters and symbols in single quotes or not in the annotations, and apply this consistently.

[rjyounes]

Since these are all one issue and a set of interrelated changes, there should be one primary bullet point and sub-bullets.

[uscholdm]

Good idea.

[uscholdm]

Recommended changes made.

[uscholdm]

@rjyounes There is a failed check that I cannot track down. I looked in the ontologyShapes.ttl and property_type_construct.rq files and found no message text that matched. The property, symbolUnicode is very much like symbol and symbolHtml. Could it be tripping over the Unicode symbols? That's the only thing that seems different enough to cause a problem.

[rjyounes]

I noticed that the rdfs:domain assertions didn't get removed from the new predicates.

[rjyounes]

Suggested change

rdfs:domain gist:UnitOfMeasure ;

[rjyounes]

Suggested change

rdfs:domain gist:UnitOfMeasure ;

[rjyounes]

Suggested change

rdfs:domain gist:UnitOfMeasure ;

[rjyounes]

@uscholdm The new symbol predicates don't conform to this shape:

gshapes:PropertyShape
	a sh:NodeShape ;
	skos:prefLabel "Property Shape"^^xsd:string ;
	sh:or (
		gshapes:LowerCase
		gshapes:NonConformingLabel
	) ;
	sh:property gshapes:MandatoryDefinition ;
	sh:targetClass
		owl:AnnotationProperty ,
		owl:DatatypeProperty ,
		owl:ObjectProperty
		;
	.

You should put these into gistValidationAnnotations.ttl file - you'll see an example there.

BTW, I also noticed that the rdfs:domain assertions haven't yet been removed from the new predicates.

[uscholdm]

The new symbol predicates don't conform to this shape:

gshapes:PropertyShape
	a sh:NodeShape ;
	skos:prefLabel "Property Shape"^^xsd:string ;
	sh:or (
		gshapes:LowerCase
		gshapes:NonConformingLabel
	) ;
	sh:property gshapes:MandatoryDefinition ;
	sh:targetClass
		owl:AnnotationProperty ,
		owl:DatatypeProperty ,
		owl:ObjectProperty
		;
	.

The property mentioned in the error message as being invalid is:

gist:symbolUnicode
	a owl:DatatypeProperty ;
	rdfs:range xsd:string ;
	skos:definition "The Unicode symbol for something."^^xsd:string ;
	skos:example "For square meter (ASCII 'm^2'), the Unicode symbol for 'm' followed by the Unicode symbol for superscript 2, i.e., 'U+006D U+00B2'"^^xsd:string ;
	skos:prefLabel "symbol Unicode"^^xsd:string ;
	gist:domainIncludes gist:UnitOfMeasure ;
	.

The label is not non-conforming. I tested it against the regex for gshapes:LowerCase at Regex101 (see below). So, the sh:or clause is obeyed. There is also a skos:definition, so gshapes:MandatoryDefinition is obeyed. I am still not seeing the problem.

[rjyounes]

The regex in the gshapes:LowerCase constraint is:

^([a-z]+|[A-Z][A-Z]+)([- ]([a-z]+|[A-Z][A-Z]+|[0-9]+))*$

^ - beginning of string
[a-z]+| - 1 or more lowercase letters or
[A-Z][A-Z]+ - 1 or more uppercase letters (I.e., an acronym like UNRWA)

([- ]([a-z]+|[A-Z][A-Z]+|[0-9]+))*$ =
0 or more of: (space or hyphen plus (1 or more lowercase letters or 1 or more uppercase letters or one or more digits))
This is meant to accommodate things like "ISBN-11"

$ - end of string

"symbol Unicode" doesn't conform. "symbol HTML" does conform.

In your regex tool, doesn't "no match" mean the input string does not match the regex? Which is the result we expect.

[uscholdm]

Thanks for that explanation. I misread the regex. Is the regex really what we want? Why rule out capitalized words, they are not that unusual.

[rjyounes]

@uscholdm The PR has been approved but has conflicts with develop that must be resolved before merging.