Thanks Andrew but I know and see the pitfalls. The use case is only 1 level deep and to be treated not as a Category but instead as simply something "Related" or "Relative" to the topic. Keeping within those 2 constraints from my tests seems to work reasonably well. Beyond those constraints... I agree, its the wild west and not something that we are interested in.
On Mon, Oct 10, 2016 at 5:20 AM Andrew Gray <andrew.g...@dunelm.org.uk> wrote: > Hi Thad, > > One quick red flag - I'm not sure how familiar you are with the > category system, but automatically parsing it without sanity-checking > can very quickly lead you into a minefield. There are a substantial > number of category trees which seem reasonable at first, but rapidly > go in unexpected directions. > > For example, if you start at "Category:Road transport" on enwiki and > go two categories deep, you get "Category:Parking facilities" - so > far, so good - but you also get "Category:The Hitchhiker's Guide to > the Galaxy", "Category:Songs about buses", and "Category:Cycling > journalists". > > There are a few pairs of categories which contain each other directly, > and a much larger number which contain each other once you go a few > levels of subcategories deep & so go in endless loops. And, of course, > a clean category tree in French might be a mess in German, or vice > versa. > > None of this is to say "don't do it", but rather "don't expect it to > be clean and tidy" :-) > > Andrew. > > On 7 October 2016 at 02:08, Thad Guidry <thadgui...@gmail.com> wrote: > > Thanks Stas, > > > > So, hmm...we'd have to build our own parser or something is what your > saying > > ? > > Because Wikidata doesn't have those kinds of connections in its graph and > > also doesn't have a SPARQL service yet against the Wikipedia > > API:Categorymembers https://www.mediawiki.org/wiki/API:Categorymembers > to > > deduce those subcategories, right ? > > > > How difficult is it for someone to create a service like the > LabelService , > > but instead using the WP Categorymembers API ? Or do you have some other > > ideas ? > > > > > > On Thu, Oct 6, 2016 at 6:10 PM Stas Malyshev <smalys...@wikimedia.org> > > wrote: > >> > >> Hi! > >> > >> > Can it all be done in SPARQL against some services that already expose > >> > WP subcategories given a specific category ? Or is there an API that > >> > does this already ? other tools that might expose WP categories ? > >> > >> I don't think subcategory relationship is not recorded in Wikidata. E.g. > >> https://www.wikidata.org/wiki/Q7361750 contains > >> https://www.wikidata.org/wiki/Q14436424 but neither have any indication > >> of that. > >> The problem I guess is that category hierarchy is different on all > >> wikis, so it's hard to have one property that expresses it on Wikidata. > >> > >> You could do through "subclass of" and "category's main topic" but not > >> sure that'd capture all. E.g.: http://tinyurl.com/h7qpcdn but that only > >> captures one subcategory, since other items don't have the same > >> hierarchy in Wikidata. > >> -- > >> Stas Malyshev > >> smalys...@wikimedia.org > >> > >> _______________________________________________ > >> Wikidata mailing list > >> Wikidata@lists.wikimedia.org > >> https://lists.wikimedia.org/mailman/listinfo/wikidata > > > > > > _______________________________________________ > > Wikidata mailing list > > Wikidata@lists.wikimedia.org > > https://lists.wikimedia.org/mailman/listinfo/wikidata > > > > > > -- > - Andrew Gray > andrew.g...@dunelm.org.uk > > _______________________________________________ > Wikidata mailing list > Wikidata@lists.wikimedia.org > https://lists.wikimedia.org/mailman/listinfo/wikidata >
_______________________________________________ Wikidata mailing list Wikidata@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikidata
