Wikidata:Property proposal/data.gouv.fr ID

data.gouv.fr ID edit

Originally proposed at Wikidata:Property proposal/Authority control

Done: data.gouv.fr organisation ID (P3206) (Talk and documentation)

Description	identifier of an organization on the French national open-data platform data.gouv.fr
Represents	data.gouv.fr (Q3016893)
Data type	External identifier
Domain	organization (Q43229)
Allowed values	letters, numbers and hyphens
Example	OpenStreetMap (Q936) → openstreetmap
Formatter URL	https://www.data.gouv.fr/organizations/$1

Motivation

This website provides a lot of free data that Wikidata can reuse. Each organization that gives data to the website data.gouv.fr has an identifier like this. Tubezlob (🙋) 14:11, 3 September 2016 (UTC)[reply]

Discussion

Comment On this page, we can download the list of all organizations (796 items) by clicking on the orange button. In this CSV file, there is an other identifier (id in the file) that the one with letters (slug), I don't know why. With OpenStreetMap, it's : 534fffa0a3a7292c64a7806b. Which identifier should we use? Tubezlob (🙋) 14:46, 3 September 2016 (UTC)[reply]

I think it's better to add the normal identifier (the one in the request) because it's easier to found and I don't understand what is the interest of a very complex identifier. Tubezlob (🙋) 22:14, 3 September 2016 (UTC)[reply]

Comment VIGNERON asked me to study the conversion between the identifiers and the textual names (slugs). I didn’t find the exact function between but I’m quite confident there is a (bijective) link between the two. When we download the file of organisations and we sort on the 'id' column, we see there is a first part with a common prefix (534fff) with names in alphabetical order and a second part with other prefixes with names without order. So it is clearly not a hash, but possibly the id are constructed on some "group" [of organisations] (see [1]). Seb35 (talk) 18:42, 8 September 2016 (UTC)[reply]

@VIGNERON, Seb35, YULdigitalpreservation: So, I change the proposal with id? It's easy to found in the source code of the page (search data-organization-id=). Tubezlob (🙋) 11:39, 18 September 2016 (UTC)[reply]

Support I prefer the textual identifiers (easily readable and more human-friendly) but maybe the hexadecimal id is better for constraint checking (allways 24 character). @Tubezlob: did you know what (hash?) function is used to generate this hexa id? Cdlt, VIGNERON (talk) 17:52, 6 September 2016 (UTC)[reply]

@VIGNERON: No idea, I have no knowledge of hash functions and things like that. But I agree with you, textual identifier seems better for our use on Wikidata. Tubezlob (🙋) 18:04, 6 September 2016 (UTC)[reply]

@VIGNERON: I've asked them on Twitter, let's wait for an answer. I'm still undecided: textual id seems better for humans, not entirely sure for Wikidata (is Wikidata for humans? :D ) ; I think it will depend on datagouv aswer and if the function is simple or not. Cdlt, VIGNERON (talk) 10:51, 7 September 2016 (UTC)[reply]

Support Given my comment above, I support the creation of such a link between Wikidata and data.gouv.fr. On the question ids vs slugs, I am more confident in the sustainability of ids, but given there is probably a bijective function between the two (with some config file somewhere), I don’t have a strong opinion on the question. Seb35 (talk) 18:42, 8 September 2016 (UTC)[reply]
Support. An external id to describe these organizations as data contributors seems useful to me. YULdigitalpreservation (talk) 14:37, 14 September 2016 (UTC)[reply]

@Tubezlob, VIGNERON, Seb35, YULdigitalpreservation: Done ChristianKl (talk) 17:19, 23 September 2016 (UTC)[reply]