Wikidata:Property proposal/WordLift ID

From Wikidata
Jump to navigation Jump to search

WordLift ID[edit]

Return to Wikidata:Property proposal/Authority control

   Under discussion
DescriptionWordLift ID, a Linked Open Data permanent ID of publicly available entities curated by online publishers and distributed in various formats.
Data typeURL
Example 1WordLift (Q31998763)http://data.wordlift.io/wl0215/entity/wordlift
Example 2Salzburg (Q43325)http://open.salzburgerland.com/de/entity/salzburgerland
Example 3cryptocurrency (Q13479982)http://data.thenextweb.com/tnw/entity/cryptocurrency
Example 4bilingualism (Q10779529)http://data.wordlift.io/wl0472/entity/bilinguismo_2
Example 5Audible.com (Q366651)http://data.wordlift.io/wl0826/entity/audible
Example 6Microsoft (Q2283)http://data.windowsreport.com/windowsreport/entity/microsoft
Number of IDs in source107,910

Motivation[edit]

WordLift ID refers to 5 stars linked data with permanent URIs publicly available online. WordLift's datasets are also published on the LOD Cloud and interlinked with other public datasets. Devbug (talk) 10:01, 6 December 2018 (UTC)

Discussion[edit]

  • Symbol support vote.svg Support - Cyberandy 11:21, 06 December 2018 (UTC)
  • Symbol support vote.svg Support Looks useful enough to me - Edei 14:10, 06 December 2018 (UTC)
  • Symbol support vote.svg Support Good and useful Content - Mreichh 14:22, 06 December 2018 (UTC)
  • Symbol support vote.svg Support Positive - Gencuo
  • Pictogram voting comment.svg Comment @Devbug: Your examples are incomplete - properties relate a wikidata item to a value, so what are the wikidata items associated with the URL values you've listed in your examples? ArthurPSmith (talk) 19:03, 6 December 2018 (UTC)
    • @ArthurPSmith: The property expresses the value of the permanent URI of equivalent entities according to the Linked Data principles, e.g. http://open.salzburgerland.com/de/entity/salzburgerland is the ID of the Salzburgerland region in Austria same as 2766823 on GeoNames or Q43325 on Wikidata. Maybe external-id is more appropriate as type and format value can be set to URL. --Devbug (talk) 19:55, 6 December 2018 (UTC)
    • I've updated the proposal to external-id with allowed values of URL. --Devbug (talk) 20:21, 6 December 2018 (UTC)
      • Well, I think URL was fine as a datatype here. What's missing are the QID's associated with these URL's - please add them to the examples. Look at other property proposals to see how it's done. ArthurPSmith (talk) 01:54, 7 December 2018 (UTC)
        • Got it, I reverted to url data type and added the QID's. --Devbug (talk) 09:45, 7 December 2018 (UTC)
  • Symbol support vote.svg Support David (talk) 09:32, 7 December 2018 (UTC)
  • Symbol support vote.svg Support This can be very useful - Domus.aurea999 10:41, 07 December 2018 (UTC)
  • Symbol support vote.svg Support Looks great to me Mark 14:27, 07 December 2018 (UTC)
  • Symbol support vote.svg Support Would be a great addition Julian 15:41, 07 December 2018 (UTC)
  • Questions: How is this different from a generic linked data URI? Is this part of a proprietary system? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 21:03, 11 December 2018 (UTC)
  • Pictogram voting comment.svg Comment @Pigsonthewing, ArthurPSmith, ديفيد عادل وهبة خليل 2: I've a strong feeling that some of the support votes above might be (sock/meat)puppets, given that their votes here are the only contributions of theirs on Wikidata and the comments attached to them seem a bit sketchy. Mahir256 (talk) 07:54, 12 December 2018 (UTC)
    • @Mahir256: I myself am David Riccitelli (Q32000705). I believe @Multichill: can vouch for me, having met at SEMANTiCS 2018 (Q50349922) where he introduced me to the property proposal process. --Devbug (talk) 10:35, 12 December 2018 (UTC)
    • @Mahir256: I expect there was some canvassing at least. However, good linked data tools are a good thing in principle. According to en:WordLift this is from a company founded a little less than 2 years ago. @Devbug: Can you describe how this approach differs from other WordPress solutions, for example the PoolParty plugin? Asserting that a given entity only has one linked data URL seems to require some sort of centralized management, no? Where is that coming from, what are the licensing rules etc? ArthurPSmith (talk) 16:27, 12 December 2018 (UTC)
@ArthurPSmith: I do think, however, that we should consider striking the votes of the two people immediately above your initial comment and the three votes right below David's support vote for being socks. Mahir256 (talk) 16:31, 12 December 2018 (UTC)
@ArthurPSmith: WordLift (Q31998763) started in 2011 within the IKS project of the European Framework Program 7 (FP7). It officially opened to the public in 2017. WordLift is the only solution for WordPress (Q13166) (as far as I know) that fully complies with the Linked data principles and the 5 stars of Linked data. In fact datasets are listed in the LOD Cloud diagram. Entity management is decentralized and happens within WordPress (Q13166), structured data is pushed to Apache Marmotta. WordLift provides also interlinking with other datasets (including but not limited to Wikidata (Q2013), DBpedia (Q465), GeoNames (Q830106), ..., by means of owl:sameAs and schema:sameAs) which is the 5th rule of Linked data "Link your data to other people's data to provide context" (and a requirement to be listed among the 1,231 datasets of the LOD Cloud). --Devbug (talk) 17:12, 12 December 2018 (UTC)
@Devbug: Not sure what you mean by "deceentralized" here. There's one Marmotta installation that WordLift is using, right? So every valid URI must be listed in that central location? Anyway, it sounds like (given the "owl:sameAs" comment) you must allow multiple URI's for the same entity, so it's not really an ID either, is it? ArthurPSmith (talk) 20:07, 12 December 2018 (UTC)
@ArthurPSmith: By decentralized I mean that the actual content is managed and stored in various WordPress instances as semi-structured data (title, content, meta fields). Because WordPress is unable to provide a performant and effective triple store, we copy the contents in the form of triples to Marmotta (which may provide also additional features, e.g. SPARQL, ldpath). I am not sure I understand the question about the ID, I'll try to give an example: http://open.salzburgerland.com/de/entity/salzburgerland is the ID for Salzburgerland like Salzburg (Q43325) in Wikidata, 2766823 in GeoNames, http://dbpedia.org/resource/Salzburg_(state) in DBpedia. Wikidata uses GeoNames ID (P1566) to state GeoNames ID and GeoNames uses the pseudo language code "wkdt" to state Wikidata's QID. --Devbug (talk) 20:49, 12 December 2018 (UTC)
@Devbug: Sorry, "not an ID" is not the right way to express what I was thinking. But just to be clear, for your Salzburg (Q43325) example there would be at least 2 (salzburgland and dbpedia) and maybe 3 (including geonames) or more (?) correct values for this proposed "WordLift ID"? ArthurPSmith (talk) 20:55, 12 December 2018 (UTC)
@ArthurPSmith: Ah, no, the correct value for "WordLift ID" would be http://open.salzburgerland.com/de/entity/salzburgerland just like 2766823 is for GeoNames ID (P1566), Salzburg-state is for Quora topic ID (P3417), etc. --Devbug (talk) 21:03, 12 December 2018 (UTC)
And the central service is deciding what those special URI's are, ok. Is there any mechanism to confirm (a lookup service?) that somebody has set the right URI? We might want to treat this as an external ID with a formatter URL if there's something that works for that... ArthurPSmith (talk) 23:28, 12 December 2018 (UTC)
@ArthurPSmith: Currently there's no lookup service, but we could provide several, e.g. one that validates a URI, one that autocompletes a URI, ... can you point me to examples of other lookup services? Initially I looked at the formatter, I am not sure it's fit, because the hostname part of the URI may be variable, i.e. by default we use http://data.wordlift.io/datasetname as base URI, however publishers can provide their own custom domain, for instance http://open.salzburgerland.com, http://data.thenextweb.com, http://dati.greenpeace.it/ and so forth. We can also prepopulate and keep the ID in sync from WordLift's side using Wikidata API. --Devbug (talk) 07:20, 13 December 2018 (UTC)
@ArthurPSmith: On behalf of Wikidata:Property_proposal/Generic should I set status=ready on the proposal ? --Devbug (talk) 09:56, 13 December 2018 (UTC)
  • @Devbug, Mahir256: I marked this as ready - I still have some curiosity about how it actually works in practice but hopefully we'll see how this is used and learn from that. ArthurPSmith (talk) 14:15, 13 December 2018 (UTC)
  • Pictogram voting question.svg Question Isn't this duplicating data we should already have? --- Jura 18:33, 13 December 2018 (UTC)
@Jura: Which property are you referring to? Can you make an example of data that would be duplicated? --Devbug (talk) 20:17, 13 December 2018 (UTC)
  • It's a question on my side. Can you give a sample of data that wouldn't be duplicated? --- Jura 09:27, 15 December 2018 (UTC)
@Jura: For example Salzburger Bauernherbst provides a description, the list of performers, the start and end dates and the relations with other entities that aren't present in Wikidata. --Devbug (talk) 15:46, 15 December 2018 (UTC)
  • @ArthurPSmith, Jura1: I am marking the property as ready - I remain available for further information. --Devbug (talk) 09:06, 18 December 2018 (UTC)
    • I think it could gain from further input from active contributors. --- Jura 09:12, 18 December 2018 (UTC)
      • @Jura: Ok, let me know if I can be of any help. --Devbug (talk) 09:20, 18 December 2018 (UTC)
      • @Jura: What kind of input are you looking for? - Edei 10:05, 18 December 2018 (UTC)
  • Symbol support vote.svg Support Very positive - fadyramzyaziz
  • @Mahir256, Multichill: I believe further comment from you here would be helpful. I don't believe Jura has grasped the purpose of this property, which is to provide the unique URI that WordLift uses (for WordPress websites) to identify concepts. There is no existing property in Wikidata that can function in this way as far as I am aware, but maybe there's an alternate way of modeling this that somebody has in mind? ArthurPSmith (talk) 14:54, 18 December 2018 (UTC)
    • You are right. Is it comparable with linking directly to the url of dublin core metadata? Not sure if any of the active contributor did either. I wonder what the impacts are and the maintenance requirements. --- Jura 16:00, 18 December 2018 (UTC)