Thad Guidry, 12/09/2016 15:46: I don't know what a "Wikidata topic" is, no such term is found on https://www.wikidata.org/wiki/Wikidata:Glossary . Your example Q474191 already has an "equivalent class" link to SchemaOrg, can you make an example of an item (Q-number) and an external thingy which you are unable to link? Thanks. Nemo

Thad Guidry, 12/09/2016 21:29: Better now? https://www.wikidata.org/w/index.php?title=Q26869695&diff=375889582&… https://www.wikidata.org/w/index.php?title=Q26869695&diff=375889582&… Do we need a help page on how hypernymy/hyponymy is handled on Wikidata? I know it can be confusing. Nemo

Hoi, I understand that this makes for more complexity inside Wikidata. That is ok if it brings us something tangible. Can you provide me with one or two practical examples how this helps? Thanks, GerardM On 12 September 2016 at 22:03, Thad Guidry <thadguidry(a)gmail.com> wrote:

Hi Andra, Yes, we'll want to have skos:closeMatch available....but also... We need to help with long-tail domain support, and one way to do that is support more 'external' properties. Wikidata doesn't have all the data in the world yet, or a full understanding of it...but many times external systems do. So I'd like to see the 'classes' properties of Wikidata to also have 'external' support. Please propose additionally, a 'external subclass' Wikidata property as well. That should do for now. Thad +ThadGuidry <https://www.google.com/+ThadGuidry>

Andra, That type of powerful expressiveness is certainly doable in Wikidata. To blatantly create properties on a whim by pseudo proxy of using Wikidata items to do the heavy lifting. But that idea is against most of what I have seen in Wikidata documentation, policy, and community best practices say to do. The right way forward is to patiently wait for the new property to exist, and then we can continue. Thank you for drafting the property proposal by the way. I'll wait for it for review. Here's some examples where we have subclasses under these : https://www.wikidata.org/wiki/Q4026292 Action should also ideally have all of Schema.org's subclasses of Action that we have like http://schema.org/TravelAction https://www.wikidata.org/wiki/Q11410 Game should also ideally have the subclass of Videogame http://schema.org/VideoGame Here's an example of flipping it around and saying that Schema.org can be used as an 'external parent class' https://www.wikidata.org/wiki/Q7889 Videogame should also ideally have a proeprty on it for 'external parent class' or something similar with the value of http://schema.org/Game Our full heirarchy is here for your perusal : http://schema.org/docs/full.html Thad +ThadGuidry <https://www.google.com/+ThadGuidry>

Thad, I actually don't see how using internal properties would go against Wikidata policy. There is the requirement of notability, " 1. It fulfills *some structural need*, for example: it is needed to make statements made in other items more useful." [1] I might be wrong, but I would consider creating a new item to describe a subclass, fulfilling a structural need. Modelling classes and subclasses this way, also makes writing federated queries where the WDQS is used in the SERVICE operator easier. I only need to consider one wikidata property to bridge Wikidata with external sources, and deal with child and parent classes in Wikidata. Cheers, Andra [1] https://www.wikidata.org/wiki/Wikidata:Notability On Tue, Sep 13, 2016 at 3:34 AM, Thad Guidry &lt;thadguidry(a)gmail.com&gt; wrote:

All, Vocabularies have their own definitions already hosted. There is no need, therefore, to replicate those same definitions in Wikidata by creating a new Wikidata item (topic / entity) for each external vocabulary class or subclass or property. Instead, the best practice is to simply POINT to those external definitions, such as those in Schema.org, DBPedia.org, MusicBrainz, etc., etc. (sorry, Andra, but unfortunately, your proposal to just create Wikidata items (recreating a vocabulary inside Wikidata, instead of using Wikidata properties to point to external URLs) makes Wikidata harder to use, not easier for itself, or for external partners or vocabularies. I won't do this, its not required, its the wrong approach, and gives grief to others that query Wikidata) The right solution is to help propose and finish adding some of the missing 'external properties' in Wikidata, then we help the web to help us. By adding those missing properties in Wikidata, as is done nearly everyday from what I see. We help external communities align with Wikidata and vice-versa. Andra - let me know once your 'external subclass' is ready for proposal review. Thad +ThadGuidry <https://www.google.com/+ThadGuidry>

The benefit is directly towards WMDE's goals. Specifically, https://meta.wikimedia.org/wiki/Grants:APG/Proposals/2015-2016_round1/Wikim… Goal 1b: Grow the reach of Wikidata beyond the Wikimedia projects A Wikidata property = Some external vocabulary property A Wikidata property is similar to Some external vocabulary property A Wikidata property is considered a parent class to Some external vocabulary class <-- Needs new property for support. A Wikidata property is considered a child class to Some external vocabulary class What Schema.org needs now from Wikidata is the addition of a few new properties that seem to be missing. (I sound like a broken record now) 1. 'external subclass'. ​DONE. Let's start there. I don't even want to drag this discussion further than beyond that 1 request, at this point in time. Thad +ThadGuidry <https://www.google.com/+ThadGuidry>

Hi Gerard, Good question. lolol. And good point, you did ask about 'how it helps' and I didn't directly answer that, so here it is. Schema.org provides a set of guidelines for publishing structured data on the web in various formats. You can read more detail on our website. The answer Part 1 : Schema.org helps publishers, small and large, even folks like you and me that might have small websites with important data, that can be crawled by bots, public scraping scripts, etc. to harvest more public data for Wikidata to absorb, as well as helping to reconcile data and even load the Primary Sources tool, Reasonator, or any tool eventually if we wanted. Adding references could in theory just be done by a bot that was smart enough to understand Schema.org markup (there are already many Python, Ruby, Java libraries that exist for this). The answer Part 2: By having Schema.org properties and classes aligned to existing Wikidata properties and classes. Both sides benefit to help understand that structured data even more....and once a machine understands it better, than those machines can be programmed to gather even more data, or validate the quality of the existing data. This mapping effort is not just limited to Schema.org however in the long term. Notice that we already have many mappings in Wikidata for external vocabularies, but its very rudimentary. We want to help Wikidata have a more full understanding of data, and that requires a few more properties to be added to support that effort. Lydia understands this. In fact, she was one of the motivators to have us encourage us to perform the mapping on both sides, inside Wikidata and inside Schema.org (both will be done, but first it was deemed necessary for Wikidata). In addition, she understands the struggle to bring in more quality data for Wikidata and to help tool authors and publishers, which in doing so helps with Goal 1b of WMDE's future plans. Lydia and a few of us from Schema.org will be having a Google Hangout to discuss a few more details this Friday. The entire effort and comments are being tracked here: https://github.com/schemaorg/schemaorg/issues/280 Thad +ThadGuidry <https://www.google.com/+ThadGuidry>

Federico, Yes they are incompatible concepts. However, one is typically treated as a subclass of the other. Its that https://www.wikidata.org/wiki/Q26869695 (nutritional diet) is a subclass of https://www.wikidata.org/wiki/Q474191 (biological needs diet) Which is correctly stated now with that 'subclass of' statement on https://www.wikidata.org/wiki/Q26869695 Yes, it looks like Wikidata did not have the nutritional concept of a Diet, but only had the biological needs concept. Thanks to your creation of the nutritional concept of a Diet, this helps everyone, Schema.org as well as dieticians, nutritionists, and other scientists. TODO: We just need to figure out what to do with 'external subclasses' and a few other things concerning Schema.org mapping. And that's part of our Agenda on this Friday's Google Hangout with Lydia and folks. Thanks again Federico for the bit of help! Thad +ThadGuidry <https://www.google.com/+ThadGuidry>