My two cents : this is a job to do in conjunction with structured wiktionary, who will be able to deal with lexical entities.
We however have some properties here and there to deal with such languages issues to deal with this inside Wikidata, female form of occupation name for example, but the logic to deal with those datas is coded in the clients like the infoboxes.
Coding this inside Wikidata would still require a step that is far from reach imho : code a per language language grammatical model that would select some lexical forms considering the context ... Not that easy to do. What would be really cool eventually is for us to code those rules in a structured way. One open question would be "how the software would know thoses rules and which one to use in which context".
I'd suggest to do this as a javascript gadget as a first step to better understand what those rule may look like, where the main units are hardcoded, and to leave the logic code in this gadget.
2016-07-29 9:19 GMT+02:00 Stas Malyshev smalyshev@wikimedia.org:
Hi!
You mean the MediaWiki message processing code? This would probably be
Yes, exactly.
powerful enough for units as well, but it works based on message strings that look a bit like MW template calls. Someone has to enter such strings for all units (and languages). This would be doable but the added power comes at the price of more difficult editing of such message strings instead of plain labels.
True. OTOH, we already have non-plain strings in the database - e.g. math formulae - so that would be another example of such strings. It's not ideal but would be a start, and maybe we can have some gadgets later to deal with it :)
Oh yes :) Russian is one, but I'm sure there are others.
Forgive my ignorance; I was not able to read the example you gave there.
Sorry, it's hard to give examples in foreign languages that would be comprehensible :) The gist of it is that Russian, as many other inflected languages, changes nouns by grammatical case, and uses different cases for different number of items (i.e. 1, 2, and 5 will use three different cases). Labels are of course in singular nominative case, which is wrong for many numbers. -- Stas Malyshev smalyshev@wikimedia.org
Wikidata mailing list Wikidata@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikidata