Re: Use AI but complement it with structured data

Hey,

Pe luni, 17 august 2026, Gerard Meijssen <[email protected]> a
scris:

> Hoi,
> Sorry for responding late, I had a wonderful holiday with my wife.
>

> I have read what I find in the link for Wikidata MCP. It brings me
> questions but no understanding. The OCLC story made sense to me because it
> described a practical application. The method includes AI, it inspired me
> to consider a potential similarity to our systems and your response implies
> that it is realistic because of the technology that is available in
> Wikidata MCP..
>
> So when we have the technology what is stopping us from actually applying
> this?
>

If by "us", you mean the WMF, I would speculate there is a mix of not
wanting to upset purists with not having resources for a truly scalable
use-case and the general slow speed of development. So, the same old
non-reasons.

If by "us", you mean you, me and the rest of the community: nothing. If
whoever is reading this email has a good idea that involves AI and linked
data, they should absolutely try it! If you need help, just shout: there
are a lot of very cool people already working with or around AI who would
love to help or point you in the right direction.

For me personally, I don't yet have that cool idea that couldn't be done
without AI, but I'm sure I'll get one eventually. And having the
infrastructure in place to use Wikidata is a great reassurance.

Best,
  Strainu

>
> I may have secondary questions, but when it helps us achieve goals that we
> can agree upon in a specific application, what is holding us back?
> Thanks,
>         Gerard
>
> On Sat, 8 Aug 2026 at 21:08, Strainu <[email protected]> wrote:
>
>> Hey Gerard,
>>
>> Have you seen the Wikidata MCP server [2]? It does exactly what you
>> describe.
>>
>> BR,
>>   Strainu
>>
>> [2] https://www.wikidata.org/wiki/Wikidata:MCP
>>
>> În sâm., 8 aug. 2026 la 12:14, Gerard Meijssen via Wikimedia-l <
>> [email protected]> a scris:
>>
>>> Hoi,
>>> I read the most wonderful blogpost by the OCLC [1]. For those that do
>>> not know, OCLC is a longtime partner of the WMF, it is the worldwide
>>> organisation of library organisations, it publishes the VIAF identifiers
>>> and they are heavily used in our projects.
>>>
>>> When you consider the OCLC blogpost, we are in a wonderful position that
>>> we have both structured data in list, categories and info boxes in our
>>> Wikipedias but also in Wikidata. In addition there is the texts that will
>>> typically confirm what it says in the structured data. Having this data
>>> enables us to find issues in our data.
>>>
>>> What I find particularly interesting is that we support more than 300
>>> languages, our AI can be trained AND functional by inferring information
>>> from our structured data. This means that we may enrich all our projects
>>> particularly where statements have weight because of the references in one
>>> language. This will materially improve our inclusivity and our relevance.
>>> Thanks,
>>>       GerardM
>>>
>>> [1]  AI needs context. Linked data provides.
>>> <https://blog.oclc.org/next/ai-needs-context-linked-data-provides/>
>>>
>>> _______________________________________________
>>> Wikimedia-l mailing list -- [email protected], guidelines
>>> at: https://meta.wikimedia.org/wiki/Mailing_lists/Guidelines and
>>> https://meta.wikimedia.org/wiki/Wikimedia-l
>>> Public archives at https://lists.wikimedia.org/
>>> hyperkitty/list/[email protected]/message/
>>> AC4UKLAU5JNWJQI3OSGQRKG5T3JHQ3QS/
>>> To unsubscribe send an email to wikimedia-l-leave-RusutVdil2icGmH+5r0DM0B+6BGkLq7r@public.gmane.org
>>
>>

_______________________________________________
Wikimedia-l mailing list -- [email protected], guidelines at: https://meta.wikimedia.org/wiki/Mailing_lists/Guidelines and https://meta.wikimedia.org/wiki/Wikimedia-l
Public archives at https://lists.wikimedia.org/hyperkitty/list/[email protected]/message/D2IUZWC33VZMRSCIJE44Y34LXVVHL5FN/
To unsubscribe send an email to wikimedia-l-leave-RusutVdil2icGmH+5r0DM0B+6BGkLq7r@public.gmane.org