Wikidata

Wikidata:Property proposal/Image link

< Wikidata:Property proposal
edit

Originally proposed at Wikidata:Property proposal/Generic

   Not done
DescriptionWhilst image (P18) is the main property for images, there would be a P with a URL link to an image that helps describe the element. It would help avoid licensing/copyright problem, so long as the image is stored elsewhere, not on Wikidata.
Data typeURL
Allowed valuesURL link
Example 1solar eclipse of June 12, 1981 BC (Q5692098) → https://eclipse.gsfc.nasa.gov/5MCSEmap/-1999--1900/-1980-06-12.gif
Example 2Post- och Inrikes Tidningar (2nd Jan 1821) (Q19536380)→ https://sv.wikisource.org/wiki/Post-_och_Inrikes_Tidningar_1821-01-02#/media/Fil:Post-_och_Inrikes_Tidningar_logotype.png
Example 3[other example for a periodic] → https://gallica.bnf.fr/ark:/12148/bpt6k7014545/f1.highres
Expected completenessalways incomplete (Q21873886)
See also

Motivation

edit

To help source images without putting them into Commons (generally to avoid copyright pb). If there is a P18 in a Q, then that new property should'nt exist. Bouzinac (talk) 07:44, 6 July 2020 (UTC)[reply]

Discussion

edit

@Pigsonthewing, Pintoch, Jura1, ديفيد عادل وهبة خليل 2, Strakhov: @Pasleim, ShadessKB, NandanaM: as a participant in this discussion, have you since changed your mind? Pamputt (talk) 21:42, 23 August 2020 (UTC)[reply]

  Strong oppose I find it weird to have the same property proposal every few years. (2014,2017, 2018) Nothing has changed in principal since the first proposal in 2014. There is no (legal) use for links to unfree pictures, they are harder to maintain than commons links (URLs change, images go offline) and will therefore require a lot of maintenance while providing no improvement. The proposed property would also violate what I perceive as the philosophy of Wikidata and Wikimedia: Spreading free data and knowledge. It is also not clear which image would be selected. There are a lot of commercial images available for most topics (and a lot of agencies that try to push their content). This might cause edit wars over the value of this property and spam edits/items for (self-)promotion of images. About ½ the RfDs i create are because of SEO-Spam. This property could make the problem of SEO-Spam even bigger. All this has already been explained by other users in previous discussions and still this proposal does not discuss these points of view. Why? If I'd create a proposal, that has already been declined several times, I'd put a lot of effort in disproving previous arguments against the proposal.

tl;drThis proposal is not practicable, not beneficial and against wikimedia principles -- Dr.üsenfieber (talk) 08:52, 12 September 2020 (UTC)[reply]

Can you clarify on why this is needed and how the benefits would be bigger then the additional effort? Concerning your straw man: I didn't say that they are impossible to maintain, I said they are harder to maintain then commons-links. Using web-links as sources indeed requires a lot of maintenance effort and will decrease the verifiability of statements if not properly maintained. However, reference URL (P854) is still useful, as it increases the verifiability on a larger scale. Allowing links to proprietary images without any constraints would decrease the overall referential integrity, and I fail to see any benefits from this proposal. -- Dr.üsenfieber (talk) 10:43, 14 September 2020 (UTC)[reply]
I'd say the use would for a reader to be inside, for instance, into that item Q5692098 and the reader wishes to see the image result. As the image https://eclipse.gsfc.nasa.gov/5MCSEmap/-1999--1900/-1980-06-12.gif is not very clear to be licence-free, it would be stored only as a link to the website (if the website show the image to any reader without login, then it's semi-free). Then the reader could confront (within few clicks) and verify information between Wikidata and that website. Obviously, if there is already an image (P18) then, there shouldn't be any image link property. It would thus help, in fine, wikidata data quality...
Anyway, I'd like to understand the seo-spam links. Could you give an exemple so that I can understand your spam-correcting problem ? Bouzinac 💬●✒️●💛 13:05, 14 September 2020 (UTC)[reply]
As far as I see, your example image is not copyrighted (source, also see Wikipedia). You say that it would increase the data quality in Wikidata, but you didn't write how it would do so. It would sure increase the quantity, but it decreases the reusability and the referential integrity. Can you please make clear, how this would increase the quality? Can you also take up the points that have been made in previous discussions? i.e. How does it relate to our goal of building a free ontology (free as in freedom, not free as in free beer)? How do you plan to guarantee referential integrity?
Well, I don't remember but I searched and was prevented from downloading NASA stuff, uploading them on Commons. So I might be wrong but I believe(d) these images were unfree. Other examples can be also front cover of newspaper. https://gallica.bnf.fr/ark:/12148/bpt6k250185j.item could have been an image link to document the printed édition of L'Humanité (Q1137404) of 8th April 1904. Gallica is an entirely free library but difficult to download client-side/upload commons/indexing. An image link would help semi-automate that task and populate Wikidata with, for instance, newspaper editions.
Anyway, about SEO-Spam: e.g. Q99306327. This item has only been created to promote the website and causes workload for the user that requests deletion and for the executing administrator. If we'd allow this property, I'm afraid that there might be more spam from providers of commercial content.--Dr.üsenfieber (talk) 14:30, 14 September 2020 (UTC)[reply]
Althought this item is obviously commercial, other items are commercial too ==> Q131005#P856 has very commercial sites too. I agree this Q99306327 item looks very obscure, but who cares ? I might be naive but no one is obliged to visit its link. Bouzinac 💬●✒️●💛 18:52, 14 September 2020 (UTC)[reply]

Unfortunately the workaround proposed by Tinker Bell does not work neither with OpenRefine nor with Quickstatements, who have it hard to handle somevalue statements. Henceforth that URL image link property would have been useful. Even if manual typing in the somevalue+URL qualifyer is possible, it is not possible to industrialize. Step to reproduce: Q5693122 P18 somevalue P2699 "https://eclipse.gsfc.nasa.gov/5MCSEmap/-1999--1900/-1999-12-05.gif" Bouzinac 💬●✒️●💛 22:00, 4 December 2020 (UTC)[reply]

Retrieved from "https://www.wikidata.org/w/index.php?title=Wikidata:Property_proposal/Image_link&oldid=1318139315"