rework or delete sports as classes (HorseRiding, Boxing, etc)

[VladimirAlexiev]

Searching for "horseriding" and "horse riding" finds only OntologyClass:HorseRiding and OntologyProperty:HorseRidingDiscipline, with no uses.

Searching for "boxing" finds more: OntologyClass:Boxing, OntologyProperty:BoxerStyle, OntologyClass:BoxingCategory, OntologyClass:BoxingStyle.

These are not used In Mapping sl:Infopolje Športna liga and Mapping tr:Futbol ligi bilgi kutusu:

 {{Condition
    | templateProperty = sport
    | operator = contains
    | value = Boxing
    | mapping = {{TemplateMapping
                | mapToClass = BoxingLeague}}}}

sport="Boxing" is mapped to class BoxingLeague (a league is a set of teams, not a set of styles).

"boxerStyle" is used in Mapping fr:Infobox Boxeur:

{{PropertyMapping | templateProperty = style | ontologyProperty = boxerStyle }}

If someone would use rdfs:range reasoning over DBpedia, that would deduce all targets of boxerStyle to be of type BoxerStyle and therefore of type Boxing.

But rdfs:range reasoning is unfeasible at present since the object property extractor doesn't pay attention to range, so it would make tons of junk. You can already see this at http://fr.dbpedia.org/sparql with query

select * {?x dbpedia-owl:boxerStyle ?y}

For http://fr.dbpedia.org/resource/Buddy_Baer it returns Martinez_(Californie) and Californie, but this city and state are definitely not boxing styles. See https://fr.wikipedia.org/wiki/Buddy_Baer, the infobox has

Style = [[Martinez]], [[Californie]]

Until and unless we have an explicit list of horse riding disciplines and boxing disciplines in Wikipedia, treating HorseRiding and Boxing as classes is dangerous and misleading. Until you have an explicit answer to the question "what are the members of HorseRiding or Boxing", don't make these into classes.

Furthermore, OntologyClass:BoxingCategory and OntologyClass:BoxingStyle demonstrate lack of ontological thinking: only OntologyClass:Boxing is needed.

[VladimirAlexiev]

OntologyProperty:BoxerStyle should be kept, since plain "style" has domain Artist (i.e "style" really means "artistStyle")

[mgns]

I removed the classes mentioned here along some other cleanup of the ontology.