some cats are not skos:Concept

[VladimirAlexiev]

Some cats on dbpedia.org don't have rdf:type skos:Concept. I discovered this while investigating skos:related. Eg try this query

PREFIX skos: <http://www.w3.org/2004/02/skos/core#>
select * {?x skos:related ?y.
  #filter not exists{?x a skos:Concept}.
  filter not exists{?y a skos:Concept}
} limit 200

[jimkont]

We need a close look at few examples but ?y is extracted from ?x and ?y page might not exist at the extraction time

[VladimirAlexiev]

Guess you're right. Added more restrictions and it now returns only a few (50ish)

PREFIX skos: <http://www.w3.org/2004/02/skos/core#>
select * {?x skos:related ?y.
  filter (not exists{?x a skos:Concept} && exists {?x ?px ?x1} ||
          not exists{?y a skos:Concept} && exists {?y ?py ?y1})
} 

[jimkont]

For example the following category is a deleted page in Wikipedia so I guess this is due to Wikipedia https://en.wikipedia.org/wiki/Category:Fictional close?

[VladimirAlexiev]

But how about eg these results of my query: http://dbpedia.org/resource/Category:Wars_involving_the_Republic_of_China, http://dbpedia.org/resource/Category:Wars_involving_Qing_Dynasty http://dbpedia.org/resource/Category:Wars_involving_the_Ming_dynasty, http://dbpedia.org/resource/Category:Wars_involving_Qing_Dynasty

I checked the latter: has only sameAs and skos:related.

• on https://en.wikipedia.org/wiki/Category:Wars_involving_the_Republic_of_China it's a red link
• on https://en.wikipedia.org/wiki/Category:Wars_involving_the_Ming_dynasty you can see that the correct spelling is https://en.wikipedia.org/wiki/Category:Wars_involving_the_Qing_dynasty (with "the").

I'd guess the reason for "page might not exist at the extraction time" is:

• deleted categories, with dangling references to them
• never-created categories, with dangling references to them

Dangling references (redlink categories) are not supposed to exist. There's an editorial policy: a category should first be created before being referenced. So these are editorial problems in Wikipedia. (That applies to both queries, not just the second one).

The extractor makes useful triples for some redlink articles (eg Person, Football_player, name from a team roster even for non-existing players in the roster).

But non-existing categories should be reported to Wikipedia, rather than producing triples (which are not useful). The question is whether you have a channel for reporting problems to Wikipedia...

[jimkont]

We don't but we want to create one. This would be a good start I guess. @ enyone has any ideas how to bootstrap this. Otherwise we can post this on the dbpedia-discussion / wikidata list

[jimkont]

@pigsonthewing can you suggest a way to report this to Wikipedia?