"beforehand owned by" & "afterwards owned by"
[edit]
beforehand owned by
[edit]
Originally proposed at Wikidata:Property proposal/Sister projects
afterwards owned by
[edit]
Originally proposed at Wikidata:Property proposal/Sister projects
When describing provenance of artworks one of the most useful properties on Wikidata is significant event (P793), while on Commons we model it with c:Template:ProvenanceEvent template. One challenge with P793 property is that different types of events require different set of qualifiers to fully describe the event. Some events can be well modeled with the existing properties (like bequest (Q211557) or art theft (Q1756454)), however the most often used group of events which describe some form of change of ownership (for example change of ownership (Q14903979), significant event (P793), sales (Q194189), auction (Q177923), purchasing (Q1369832), trade (Q601401), etc.) do not have usable properties for describing old and new owners. As a result, many artworks are missing this most basic information about the event, or items use badly fitting existing properties resulting in unclear statements open to interpretations.
Some examples of such confusing or statements:
- At Mona Lisa (Q12418)
where "of (P642)" property is used to specify either seller or the buyer. In above case it is a buyer, but at almost identical statement for The meeting of Jacob and Rachel (Q53673495) it is a buyer. I am not sure of other languages but in English it seems like we are talking about "acquisition" "of" that person.
- At Girl with a Pearl Earring (Q185372)
where owned by (P127) property is used to specify the owner either before or after the transaction, but it is not clear which one.
So the motivation for those properties is to
- be able to accurately models statements like "Acquisition method: purchased from Thomas Agnew and Sons Ltd, 1937"[1], which is currently not possible
- to fix previously added ambiguous statements
- to provide better mapping between Wikidata's data model and the way we model the same data since 2009 using c:Template:ProvenanceEvent on Commons and allow better use of Wikidata resources by c:Module:Artwork and other projects
--Jarekt (talk) 05:22, 14 February 2022 (UTC)[reply]
Support A question might be: how much of this is redundant to owned by (P127), or could be done via that property. But I think these two new properties would be much neater, notably for modelling 'provenance events' where we have a particular pattern of data.
Yes, a "beforehand owner" should also be indicated by owned by (P127) + end time (P582) + end cause (P1534), and some people won't like that redundancy. But IMO, when we have data about a particular sale event, it is useful to be able to pull that together on a single statement, together perhaps with additional information like venue or auctioneer. Yes, perhaps one might be able to infer from "X owned the portrait until ..." + "Y owned the portrait from ...", that X had sold it to Y. But that might be an inference that might be wrong, because the actual chain might have been, X sold it to dealer A, who sold it at auction to dealer B, who sold it privately to Y.
So having the information of the actual transfer in a single significant event (P793) statement is useful, I think, and does seem more secure and a more sure way of modelling things, than trying to draw inferences from owned by (P127) statements even when apparently successive. I therefore support these two properties, as together creating a useful neat and tidy model for such transaction events. Jheald (talk) 16:00, 14 February 2022 (UTC)[reply]
Comment Provenance is complicated so thanks for adding these proposals and spelling out issues with Commons templates that I heavily use but which are still a mystery to me. I suspect this conversation can better be had over at Wikidata:WikiProject Provenance, which I set up to gather such questions and answers. Despite the desire to place provenance events in a distinct timeline, I doubt it’s always possible to do with “before & after” context in a verifiable way that would make such properties useful. In the end we would like to use dates as much as possible but for inheritance chains it would be nice to be able to do those “by inheritance” handoffs semi-automatically. Jane023 (talk) 15:50, 15 February 2022 (UTC)[reply]
- Jane023, Thanks for alerting me about Wikidata:WikiProject Provenance. I did a lot of digging about current providence modeling, but never run into it. --Jarekt (talk) 02:58, 20 February 2022 (UTC)[reply]
- I had to dig a bit to find Wikidata_talk:WikiProject_sum_of_all_paintings/Archive/2020#Refining_provenance_events, nothing really happened so thanks for bringing this up again. This is a more generic problem. We have events that involve multiple actors, one persons sells and the other buys. That's both the same event. This proposal seems to be too limited in scope. Probably needs more refinement. Maybe we can have a look how this gets modeled in other place? So
Oppose for now to prevent premature creation. Multichill (talk) 21:05, 16 February 2022 (UTC)[reply]
- Multichill, I am fine with refinement if there are some other ideas. Those 2 properties were based on how we were modeling it on Commons for the last decade, but I am open to alternative proposals. One think I would like to avoid is postponing this discussion for next number of years, because people are trying to model it now and the results are incomprehensible . A lot of provenance data you can find for artworks is an array of ownerships punctuated by change of ownership events. So we can model it either by owned by (P127) with the 2 book-end events or by significant event (P793) with before and after owners. I doubt you can find other approaches, but I am OK with wider discussion on how to model this (perhaps at Wikidata:WikiProject Provenance), before we proceed. --Jarekt (talk) 02:58, 20 February 2022 (UTC)[reply]
Comment Multichill, Jarekt: To "Maybe we can have a look how this gets modeled in other place?" (repeating myself from here where Jarekt considere these properties in 2018 already): "They would be similar to CIDOC-CRM's P22 transferred title to (acquired title through) and P23 transferred title from (surrendered title through)".
- I think this is a good proposal, I think I want support it, but want to have a look into the CIDOC-CRM solution another time and read the full discussion here before. Are there more specific, good reasons against the properties, also considering the "postponing this discussion for next number of years"?? Best, --Marsupium (talk) 07:56, 30 April 2022 (UTC)[reply]
Comment same as Jane023 and Multichill. Additional to Jane023 there are not only heritage chains but also sales-chains and combined chains by sales, heritage, gift and bequest. So we have to model credit line. And we have Wikidata_talk:WikiProject sum of all paintings/Archive/2021#Provenance, continued. There is also a need to express on loan; --Oursana (talk) 12:19, 18 February 2022 (UTC)[reply]
Comment It's clear the properties of (P642) and owned by (P127) wont work well as qualifiers in the context. Other than doing away with them entirely (which nobody seems to favor), are there are any valid alternatives? If these were formulated somewhere else (onsite, offsite, wherever), please quote them here. --- Jura 17:11, 3 March 2022 (UTC)[reply]
@Jarekt, Jheald, Jura1, Jane023, Multichill: I just want to give you an example with gift, which in my opinion works fine with owned by (P127) and gift: Statue of Venus (Q98784999). "previous owner" / "subsequent owner" can also be expressed by date and series ordinal (P1545). In my example I would rather delete significant event (P793)--Oursana (talk) 19:19, 12 March 2022 (UTC)[reply]
- One of the main points was to reduce the use of "of" in significant event. In this case Oursana you have used significant event to show that Louis XV gave it as a gift in 1750. I added this the same way I have been doing for bequests, with donated by (P1028) with qualifier point in time (P585). To indicate reception, I added the qualifier start time (P580) to the relevant collection (P195) statement. It's still missing a location (P276) statement. Just saying "we need to model the credit line" is much more complex than adding a 'credit line' property or making a 'credit line' significant event. Jane023 (talk) 10:03, 13 March 2022 (UTC)[reply]
- As to the edits, yes you can now sort-of work out what happened (or at least you can as a human), by spotting that the start of the time in the collection at Minneapolis matches the time of the donation; but I do think that the "subsequent owner" property proposed here would a valuable addition, to make the chain a lot more concrete, without needing to use a certain amount of inference to 'join the dots'.
- Good move to try to get rid of of (P642) though, it's a menace. Jheald (talk) 19:56, 14 March 2022 (UTC)[reply]
- Thanks for that clip! Bouguereau annoys me personally for a multitude of reasons, but this wasn't one of them until now. Thanks for the approval and my intention was to show how modelling provenance can never be done in one property, but it touches virtually all properties we currently use for paintings (including at times, dimensions and painting surface). In this case, I also changed the significant event itself, from the general sales (Q194189) to art auction (Q74570489), because the qualifier lot number (P4775) was used. I have been using this combo for works that have sold for high prices, such as this one. That said, in my opinion the other more significant part of this event in this particular case is deaccessioning (Q25339601) which I added under has cause (P828). The reason I set up the WikiProject Provenance was to centralize decisions on this. As far as I can tell reading through this proposal, the before and after are just for use in modelling significant events for artworks? I assume because we haven't decided what types of Qids can be the object of the before & after? I would like to draw a line in the sand as far as types of entities that can be collections or owners, but have been unsuccessful thus far in ordering my thoughts on this. I am unwilling to touch P127 on the Mona Lisa, but I disagree with stuffing whole countries in there. To sum up my objection to this specific proposal: There are more ways to get rid of "of" by getting rid of the significant event using other properties. Jane023 (talk) 11:43, 15 March 2022 (UTC)[reply]
- Marsupium, I do not like using anonymous (Q4233718) item. Many issues with it were described at Wikidata:Requests for comment/Cleaning up the ontology of anonymous; However the solution is to replace anonymous (Q4233718) with "unknown value" and add object of statement has role (P3831) -> anonymous (Q4233718) qualifier, and this solution does not seem to work for anonymous (Q4233718) used by a qualifier. We could agree that "
some value
" used with those properties mean "anonymous" seller or buyer but it is not intuitive. --Jarekt (talk) 17:16, 24 May 2022 (UTC)[reply]
- Ah, yeah, I see the point that adding the qualifier is not possible. Thanks for pointing this out! --Marsupium (talk) 19:34, 15 November 2022 (UTC)[reply]
Support per all above, modelling for this is needed, the proposed properties solve at least part of the problems, I don't see huge issues with this proposal, no good alternatives shown, proposed modelling is in line with CIDOC-CRM. --Marsupium (talk) 08:42, 7 May 2022 (UTC)[reply]
- Alternative shown below. 89.14.237.54 20:31, 23 March 2023 (UTC)[reply]
Support This appears to be a valuable proposal. Would this allow "looted" or stolen works to be documented after they are recovered or rediscovered? --Ooligan (talk) 18:46, 6 October 2022 (UTC)[reply]
Comment To me this proposal looks ready for creation by now. Are there still concerns or objections? Otherwise mark this ready for creation? I think it would be quite a progress and help with proceeding recording provenance. Thanks for any comments! --Marsupium (talk) 19:39, 15 November 2022 (UTC)[reply]
Support JASHough (talk) 02:21, 5 December 2022 (UTC) Looks like a very useful addition, I have been struggling with adequately representing provenance and this would solve many issues.[reply]
Oppose create an item for each significant event instead. In the event item store e.g. "P31=trade", and participants using qualifier role = thief, seller, buyer etc. 89.14.237.54 20:27, 23 March 2023 (UTC)[reply]
Support --Gymnicus (talk) 09:13, 10 May 2023 (UTC)[reply]
Done beforehand owned by (P11811) afterward owned by (P11812) Midleading (talk) 14:33, 3 June 2023 (UTC)[reply]