User talk:Lockal/Archive/2016/1

Latest comment: 8 years ago by Jura1 in topic Thanks for cleaning up...

Wikidata weekly summary #191

De Drie Stoepen

De Drie Stoepen
Why did you do this ? When you do not motivate your action, how can one assert it's validity ? Or learn from it ? Also one can perceive this lack of common courtesy as very rude. Let's see if we can set this straight, so it will not linger around ?
Regards, --OSeveno (talk) 17:41, 17 January 2016 (UTC)

@OSeveno: The statement The Three Steps (Q21997535)significant event (P793)1901 (Q2035) makes no sense whatsoever. In technical terms, it violates the first constraint of significant event (P793): Value type "event occurrence (Q1190554), fictional occurrence (Q14136353)" (see this page). The revert was self-explanatory: 1901 is not an event. Unfortunately, Google Translate does not work for nl:De Drie Stoepen well, but as far, as I understand, the correct statement could be
Note that start time (P580) and end time (P582) are added as qualifiers, not as statements (see Help:Statements). Adding correct statements is up to you and other people from nlwiki and I'm ready to help, should you have any problems. --Lockal (talk) 20:34, 17 January 2016 (UTC)
See also: Special:ConstraintReport/Q21997535. --Lockal (talk) 20:51, 17 January 2016 (UTC)

Wikidata weekly summary #192

?item

Hi Lockal,

For a sparql output to work with Listeria, currently there needs to be at least one column with labelled ?item and returning distinct items. This even if they are not needed ;)
--- Jura 08:15, 25 January 2016 (UTC)

Yep, already got this. --Lockal (talk) 08:24, 25 January 2016 (UTC)
I tried to add your idea to Module:Property documentation‎, but it times out on the more frequently used properties.
--- Jura 09:10, 25 January 2016 (UTC)

Wikidata weekly summary #193

Wikidata Graph Builder feature request

The Wikidata Graph Builder is a great tool, well done. If you're still developing it and are taking suggestions for new features, here are two that I think would be useful:

  • There could be an option to make the query go both forward and backward, instead of having to select only one.
  • Having the option to use multiple properties simultaneously (maybe with the connections different colors, labels, or shape markings?) would also be useful.

--Yair rand (talk) 01:01, 26 January 2016 (UTC)

@Yair rand: I've added "Both" directions option (example, clear js cache if you have any problem). Probably not very useful yet, but I am all into adding new options. I'll think about UI for multiple properties/colors/etc. --Lockal (talk) 21:51, 26 January 2016 (UTC)

графиня

Hi,

I see you add « графиня » on count (Q3519259) while Infovarius put the same word for graf (Q28989). These two concept are very close but in many languages there is two words for them (except in German obviously), I see that the label is « граф » (obviously for the German) in both item, could you check if it is right? If it is, please add it as an exception on Property talk:P2521 constraints.

Cdlt, VIGNERON (talk) 15:24, 26 January 2016 (UTC)

Thanks for cleaning up...

@Jura1: who also had some,

Hello,

I see that you found and cancelled more of the wrong edits I accidentally made on January 18 [3]. Thank you very much.

I had a problem with HarvestTemplate, which imported Birth dates from enwiki in P570 instead of P569. Thought everyone of these had been taken care of, by Mbch331 or by me, but obviously, there were some left.

Actually, I don't know how to find them systematically to clean them up. The report for 569=P570 does not show more of them - hope there won't be.

Sorry for giving you such overwork. :/ --Hsarrazin (talk) 17:26, 25 January 2016 (UTC)

@Hsarrazin: Actually it was a newly-created Wikidata:Database reports/items with P569 greater than P570. "P569=P570" does not handle cases where date of birth has YYYY-MM-DD precision and date of death has YYYY precision. And according to sparql, 2000 is less than 2000-04-04.
@Jura1: For now I've set this report to compare years with years. More generic queries either have too much false positives (e. g. comparing years with centuries) or fail because of timeouts. I guess the right idea for now is to create multiple reports for date-date, date-year, year-date and year-year. Unless someone creates query, which does not time out. --Lockal (talk) 19:18, 25 January 2016 (UTC)
Well, initially even the P569=P570 kept timing out. I started doing some of the cases with the same year for both, marking "infant deaths" to exclude them going forward. Personally, I think it could be worth to attempt to clean up all P569>P570 even if it means making changes like this.
It did include some of your edits, @Hsarrazin:, but given the large number of additions you made in recent days, I think it's minor. Your imports from otherwise not looked after wikis allowed to identify various duplicate items.
--- Jura 19:56, 25 January 2016 (UTC)
Thanks to this new report, I found items with errors on source wp. I put messages on their Village pump pages to keep them informed of the problem, so they can correct it (zh, uk…) — wikidata is also a powerful tool for this kind of error ;) --Hsarrazin (talk) 20:05, 25 January 2016 (UTC)
I changed the 569=570, limiting it to the year. This should allow to undo most problems easily.
--- Jura 20:47, 25 January 2016 (UTC)
\o/ thanks Jura, much more useful...
the problem of century birth date stays, like here — in these occurrences, I try to approximate the date as best as possible, and use it with century precision, and add a "before" qualifier. It avoids those "same dates"… --Hsarrazin (talk) 22:42, 25 January 2016 (UTC)
It seems there a lot of these. Maybe the approach isn't ideal - let me give it some more thought. BTW, the guy you mention is probably still alive, even if google and ruwiki says the contrary. The date is that of his predecessor.
--- Jura 07:29, 26 January 2016 (UTC)
@Jura1: - maybe you could remove from the list all persons where one or the two dates are century precise ? :) --Hsarrazin (talk) 12:46, 31 January 2016 (UTC)
There is something odd about the way they are displayed there. It uses the decade format instead of the century format.
In general, we would need to find a preferred way of setting century precise dates. I'm not sure if the current approach is ideal. It might be easier to just ignore them.
--- Jura 13:11, 31 January 2016 (UTC)
On this, can I just remove all P570 added between 23:12 and 23:28 on the 17th?
--- Jura 07:59, 26 January 2016 (UTC)
Return to the user page of "Lockal/Archive/2016/1".