>>   There was a link to disambiguation page which was detected using
>> other tool which is not using wikidata.
> Could you point me to that tool?

  It is a local Lithuanian tool. But here you can have a look at results:
  http://patrulis.openmap.lt/wikipedia.html

> That's exactly my point. I mean, that's why I think it's a good idea to
> automate the process as much as possible. Even in cases requiring human
> attention it is possible to make it easier if a QA tool gives links needed
> to decide what is correct.

  And I'm showing you examples of exactly the opposite. Most of the
errors found need fixer to READ, LOOK around and make DECISION. Even
with those "simple" ones as missing wikipedia link on a school object
needs reading, because there could be a wikipedia article about some
former school which currently is something else.
  As far as I know wikipedia has no way to specify object as
"historic" - which does not belong to OSM.

>>   It's not only names, but codes and some other details. Wikipedia
>> page content is probably mixed or swaped (haven't done analysis yet).
> That seems like a problem to fix in Wikipedia and one may assume in good
> faith that when it's solved, everything about hillfort I will be in article
> about hillfort I and analogically for hillfort II. And hillfort I will be
> the one marked as such on the ground, as it is easier to fix Wikipedia, than
> go and change information on tables in front of the object. So, I would go
> and linked hillfort I in OSM with hillfort I in Wikipedia and waited until
> Wikipedia guys fix in Wikipedia, what needs to be fixed. In fact, I have
> done so, but then I thought again and reverted it. It's your decision.

  I claim it is better if the person who knows about such stuff is
doing the changes. Not somebody applying auto-guesswork and hoping
somebody else to finish (to do actual work, and usually do MORE work).

> How were the people asked? I can only see very short note in Lithuanian. I
> can' understand it, but it doesn't seem like "do not touch" request...

  Have you noticed the title of this thread? ;-)

> My advice would be to put some note in English. If you really think it's
> needed - see above.

  It looks ridiculous that we now have to put up signs all around to
avoid guessfixers... It is much simpler to just revert and be done
with it.

> You can also opt out from my script now by just adding "nowikidata=yes".

  Would it be enough to put nowikidata=yes, noautoupdates=yes,
noguesswork=yes on the node for Lithuania as a country? ;-)

> And now, think: how the existence of wikidata-based QA tool would stop
> wikipedia-based tool from detecting the error?

  It will not stop it. Wikidata based tool is also a very good idea. I
have no doubt we will be using it as well.

  But its current implementation is unacceptable to us. We want people
to know and think, not to be dumb commiters of auto-calculated
guesswork. So removing wikidata is currently the only way to avoid
such unwanted fix-hikers. And note, we're not mass-removing wikidata
tags, only on objects which attract attention by this
wikidata-"fixer".

  And again. If you guys were asked a number of times during previous
years to STOP doing updates at least in Lithuania (and also in other
countries), what is a point of continuing doing that? You're not
helping.

-- 
Tomas

_______________________________________________
talk mailing list
talk@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk

Reply via email to