Hoi,
FORGET ABOUT REASONATOR, FORGET ABOUT WIKIPEDIA, FORGET ABOUT WIKIDATA
It is about sharing information. That is what this is all about. The
information is NOT in Wikipedia, only the data is in Wikidata, there are
plenty examples of that. Redirects are something you come up with because
it completely focuses on Wikipedia while actually it is VERY much in the
way when you want to inform people.
You do not see the problem. You do not even understand why your "solution"
is imperfect, not even halfway sane. When you forget about Wikipedia for a
moment, you will agree that Wikidata has tons of data Wikipedia does not.
Consequently, it would make sense to provide our readers with information
when Wikipedia does not have it. Wikidata is NOT informative, it takes
something like Reasonator to make the data informative.
I do not want anything less for Wikipedia.
When we approach our customers with the sum of all the information we have
available to us, you will find that Wikidata knows about something like 50%
more subjects. It impacts everything from search results, categories, red
links and disambiguation pages.From such a perspective linking "redirects"
to Wikidata is an awful idea for all the reasons I presented.
Redirects will harm Wikidata, there is no doubt in my mind. There will be
not be much of a benefit.
Thanks,
GerardM
On 22 October 2014 15:58, James Heald <j.heald(a)ucl.ac.uk> wrote:
Gerard, I still don't see a problem.
If somebody wants to search on Reasonator, they can search on Reasonator,
and they will get exactly the same Reasonator pages as before -- the only
difference is that those Reasonator pages will include more links to
relevant Wikipedia pages, with some of them badged as redirects.
As for "Death of Alice Gross", I don't see the problem there either.
Your complaint appears to be that at the moment people directly sitelink
Q(Alice Gross) to "Death of Alice Gross", causing all sorts of mismatches
and confusions.
Allowing sitelinks to redirects would actually *solve* this issue, because
then people could site-link Q(Alice Gross) to "Alice Gross" (a redirect).
Q(Alice Gross) would then no longer be sitelinked to an article about an
event; but instead would be sitelinked to a redirect.
Wouldn't that be a better state of affairs ?
-- James.
On 22/10/2014 12:19, Gerard Meijssen wrote:
Hoi,
I do not consider myself confused. I am speaking plain language.
The article: "Death of Alice Gross" has information about a living person
while it is NOT a living person. As it is, current practices like with the
"Death of Alice Gross" are problematic already enough.
When you want redirects, you make the situation worse because you will
want
to include many more people who go by a same name. Many of them are
already
known to Wikidata. We do not need redirects in Wikipedia to link to them .
What we need is integrated search where results from Wikidata and
Wikipedia
are mixed in order to provide the best result. When there is no article
about someone or something, we can provide a reasonator kinda screen with
information in English. It will refer to all kind of related information
and by having this information in Wikidata, this information is available
to any and all other languages as well.
The point is very much that any Wikipedia does not include all the
information we know about. We know in Wikidata about many more items than
Wikipedia has articles for. We can express this information in a much more
informative way than by having redirects. The examples of redirects given
were really not informative. It is not possible to associate categories
and
templates in a way that makes them useful in any other way. It positively
destroys the usability of information from Wikipedia in this way.
For what ?
We can and should do better. It starts by considering all options. Text is
no longer the only game in town.
Thanks,
GerardM
On 22 October 2014 10:03, James Heald <j.heald(a)ucl.ac.uk> wrote:
Gerard, you seem confused.
>
> (1) There would be no change to the item structure on Wikidata in any way
> -- no change to the values of any of the item properties -- only some
> extra
> sitelinks.
>
> So I don't see *why* you think there would be any risk to Wikidata's
"own
> integrity".
>
> In particular, there would be no change at all to what Reasonator would
> be
> showing, apart from a few extra badged sitelinks.
>
>
> (2) You seem to be worried that Wikidata would pick up and import the
> categories of the article that the redirect redirects to.
>
> But there's no obvious reason why this should happen. It would not be
> those articles that Wikidata would sitelink to, but the redirects. So it
> would be the categories (if any) of the redirect that would be relevant.
>
> Similarly, it would not be the item sitelinked to the redirect that any
> template on the article that was the target of the redirect would compare
> itself with -- the target article would have its own item, just as it
> does
> today; so just as it is today, that is the item that any templates on
> that
> article would compare themselves to; or that any data migration would
> load
> data into -- just exactly the same as it is today.
>
> "Death of Alice Gross" is not the article about Alice Gross.
>
> But this is not the article that would be sitelinked to Alice Gross.
>
> Instead "Alice Gross" (a redirect) is the article that would be
> sitelinked
> to Alice Gross.
>
> So none of the problems you foresee should occur.
>
>
> (3) Reasonator is great. But ultimately, Reasonator and Wikidata can
> only give a summary of the facts.
>
> In cases like Daniel Havell, and the question of his exact relationship
> to
> other members of the Havell family, Wikidata/Reasonator can note that
> sources disagree. Wikidata/Reasonator can identify a preferred value.
> But
> it is harder for them to present the context as to *why* that value is
> preferred, in the way that can be done in continuous free text.
>
> It is good to make Wikidata/Reasonator as comprehensive as possible; but
> there is added value in having the ecosystem of text Wikipedia connected
> to
> them.
>
>
> (4) One additional point is that by tracking the redirects, specifically
> by adding a property noting what items an item may redirect to in
> different
> languages, we actually improve Wikidata.
>
> * We add to the "related items" that Wikidata can display.
>
> * We make it possible to ask whether the item can be connected to these
> new additional 'related items' within one, two, three, or
''n'' hops,
> using
> the item's existing properties. If it cannot, then there is probably an
> existing property that is missing. So we can identify ways to build and
> improve the database.
>
>
> In summary: your apparent view that linking to redirects will lead to
> data being migrated onto the wrong items on Wikidata seems to me to be
> mis-founded.
>
> Instead, allowing sitelinking to redirects that accurately match the
> topic, rather than enforcing that sitelinks can only be to primary
> articles
> (which may not quite so closely match the topic), is, if anything, likely
> to create a *more* accurate structure, which will make make *less* likely
> any risk of item data pollution through ingestion from a
> not-quite-properly-matched article.
>
> (ie: if linking to redirects is supported, it will make it *less* likely
> that users will be tempted to sitelink :en:hatmaking directly to
> :d:hatmaker, and so *less* likely for data to be ingested to a wikidata
> article from a not-quite-comparable wiki article).
>
>
> I look forward to your comments.
>
> All best,
>
> James.
>
>
>
>
>
>
> On 22/10/2014 06:43, Gerard Meijssen wrote:
>
> Hoi,
>> When a position is taken that is manifestly wrong, it is worse to
>> desist.
>> Andy I like you too but calling someone a dick because he does not agree
>> with you and calls bullshit on the points taken, the examples supplied
>> is
>> not in the best tradition of our projects.
>>
>> Wikidata is NOT there to serve the English Wikipedia at the expense of
>> its
>> own integrity. A wish has been formulated to support redirects by
>> WIkipedians while Wikidata has been EXPLICITLY designed NOT to support
>> redirects but more importantly parts of articles.
>>
>> If a project does not have or want to have an article on a given
>> subject,
>> Wikidata can provide information when used in combination with the
>> Reasonator.
>>
>> Articles are about a subject and CONSEQUENTLY they should have
>> categories
>> and info boxes that are in line with the subject of the article. The
>> ARTICLE 2014 ISIL beheading incidents
>> <http://www.wikidata.org/wiki/Q17985279> for instance is NOT about a
>> human
>>
>> and it should NOT have a category "deaths in 2014" or any other
>> information
>> that is particular to one person. The same is true for "Death of Alice
>> Gross"; it is NOT about Alice Gross. When an article is just text and
>> nobody cares about such consistencies, fine. However, you want articles
>> like this linked and someone else is to clean up such mess. This
>> prevents
>> automated processes, it is bad practice and it is part of the same
>> practice/school of thought whereby we are to have redirects ... Hell
>> no!
>>
>> Please reconsider your arguments and please do not be a dick yourself..
>> Thanks,
>> GerardM
>>
>>
_______________________________________________
Wikidata-l mailing list
Wikidata-l(a)lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/wikidata-l