I wondered if anyone else is actively working on fleshing out the company data within Wikidata?
Is https://www.wikidata.org/wiki/Wikidata:WikiProject_Companies/Properties the best reference guide?
Looking to chat, on- or off-list, with interested people, hear about academic research being done, challenges you are facing, etc. Even set up a meetup if there is anyone in or near London.
As background we're organizing (paying) some people to improve the quantity and quality in a couple of areas (*), with the ulterior motive of reaching a critical mass, so the information becomes useful and other users or even the companies themselves start maintaining it.
Some of the above page is still vague, and the examples are simple. We want to be able to cope with complex business structures, joint-venture child companies, companies in multiple industries, with many brands and products, etc.
Darren
*: Not coincidentally, areas that are useful to a couple of our clients. "We" is my company, QQ Trend Ltd., a small data/AI company.
Hi Darren,
OpenStreetMap Started linking company data with wikidata ( https://github.com/osmlab/name-suggestion-index ) ( https://wiki.openstreetmap.org/wiki/Key:brand:wikidata ) current status: https://taginfo.openstreetmap.org/keys/brand:wikidata ( ~ 60000 objects ) the project is still in its early stages, Lot of issues & data modeling problems.
best, Imre
Darren Cook darren@dcook.org ezt írta (időpont: 2019. jan. 23., Sze, 11:07):
I wondered if anyone else is actively working on fleshing out the company data within Wikidata?
Is https://www.wikidata.org/wiki/Wikidata:WikiProject_Companies/Properties the best reference guide?
Looking to chat, on- or off-list, with interested people, hear about academic research being done, challenges you are facing, etc. Even set up a meetup if there is anyone in or near London.
As background we're organizing (paying) some people to improve the quantity and quality in a couple of areas (*), with the ulterior motive of reaching a critical mass, so the information becomes useful and other users or even the companies themselves start maintaining it.
Some of the above page is still vague, and the examples are simple. We want to be able to cope with complex business structures, joint-venture child companies, companies in multiple industries, with many brands and products, etc.
Darren
*: Not coincidentally, areas that are useful to a couple of our clients. "We" is my company, QQ Trend Ltd., a small data/AI company.
Wikidata mailing list Wikidata@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikidata
Thanks, that was interesting. As you find data problems, are the fixes fed back into Wikidata?
The top entry here caught my eye: https://taginfo.openstreetmap.org/keys/brand%3Awikipedia#values
I.e. 5629 Japanese 7-11 stores have a link. Apparently there are 20,392 7-11s in Japan, so that represents about 27%.
Ministop was the 2nd Japanese entry in the list, but has only 461 out of their 4902 stores linked, so just under 10%. (Family Mart has 17,400 (zero tagged); Lawson has 14,000 (127 tagged).)
That difference suggests it is not just users putting in their local store, but something more systematic? Was it done by the 7-11 company?
Darren
P.S. https://taginfo.openstreetmap.org/keys/brand%3Awikidata#values has 5788 for Q259340 (i.e. 7-11); the others are a 100 or US stores.
OpenStreetMap Started linking company data with wikidata ( https://github.com/osmlab/name-suggestion-index ) ( https://wiki.openstreetmap.org/wiki/Key:brand:wikidata ) current status: https://taginfo.openstreetmap.org/keys/brand:wikidata ( ~ 60000 objects ) the project is still in its early stages, Lot of issues & data modeling problems.
best, Imre
[Using Wikidata] we want to be able to cope with complex business structures, joint-venture child companies, companies in multiple industries, with many brands and products, etc.
https://www.wikidata.org/wiki/Q259340 is a great example of how it gets complicated.
Should Country be USA or Japan?
If instance of "chain store", should the legal form still be "kabushiki gaisha"? Or should that just be on the parent organization?
The industry of "wholesale" just seems wrong. "Retail" or "Franchise"?
Is headquarters location "Irving"? That is the HQ of "7-Eleven America", while the HQ of "Seven & I Holdings" is in Chiyoda, Tokyo.
The Facebook ID is "7ElevenMexico" which describes itself as "Retail company in San Nicolás de los Garza" ("7Eleven" is the U.S. chain; "711.SEJ" is for the Japanese shops)
I'm wondering if the problems stem from the same entry being used to describe both aspects of the brand, and aspects of the various global companies that own and use that brand.
Darren
Hi Darren,
Currently, in Wikidata, you will see a mixing of Brand and Organization. Brands can be acquired, bought and sold. Wikidata has not set forth a hard rule of "don't treat Organization as a Brand". And that's fine, I prefer keeping this a bit loose because it helps discovery sometimes when we only have the Organization. And dealing with a Brand and Ownership of a Brand or Marks can be done outside of Wikidata, where Wikidata just has good linkage to registration numbers, etc.
Ideally, some link(s) would be made between a Brand Holder in Wikidata (owner of) and Marks(registration number) which in the legal & marketing worlds culminate into a Brand (company identity). https://www.wipo.int/branddb/en/?q=%7B "searches":[{"te":"Apple%20Inc","fi":"HOL"}]}
Recently, on Schema.org we had a good discussion going about all of this: https://github.com/schemaorg/schemaorg/issues/2129
Darren Cook, 23/01/19 12:07:
I wondered if anyone else is actively working on fleshing out the company data within Wikidata?
There's also https://www.wikidata.org/wiki/Wikidata:WikiProject_Organizations#Where_to_potentially_search_for_properties.
What always puzzles me is that Wikidata has tons of details about entities but almost nothing when it comes to the basics, such as revenues/budget and number of employees, the kind of information which is most often updated in infoboxes.
Federico
In my experience, the problems highlighted here (e.g. lack of coverage, consistency, and/or accuracy) are typical of collaborative data projects such as Wikidata and Wikipedia. Using Wikidata and the likes to power entity-oriented applications automatically is always challenging. But I guess the lack of constraints is also one of the main reasons why those projects are successful :)
@Darren: can you elaborate on "I wondered if anyone else is actively working on fleshing out the company data within Wikidata?".
I guess most edits are manually made by individual editors based on personal interest.
And dataset imports are typically both small-scale and specific, like this recent project to import video games companies: https://www.wikidata.org/wiki/Wikidata:Dataset_Imports/Video_Game_Companies
Cheers. -N.
On Wed, Jan 23, 2019 at 9:27 AM Federico Leva (Nemo) nemowiki@gmail.com wrote:
Darren Cook, 23/01/19 12:07:
I wondered if anyone else is actively working on fleshing out the company data within Wikidata?
There's also < https://www.wikidata.org/wiki/Wikidata:WikiProject_Organizations#Where_to_po...
.
What always puzzles me is that Wikidata has tons of details about entities but almost nothing when it comes to the basics, such as revenues/budget and number of employees, the kind of information which is most often updated in infoboxes.
Federico
Wikidata mailing list Wikidata@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikidata