Wikidata:Property proposal/source of action & destination of action

From Wikidata
Jump to navigation Jump to search

source of action[edit]

Originally proposed at Wikidata:Property proposal/Generic

   Not done
Descriptionqualifier to be used when the object of a statement is an action, to qualify what's the source of the action
Data typeItem
Example 1cp (Q305946)has use (P366)copying (Q1156791)action applies tocomputer file (Q82753)source of actionlocal host (Q115517044)destination of actionlocal host (Q115517044)
Example 2
Example 3
  • git clone (Q115519944)has use (P366)cloning of a repository (Q115520056)action applies to git repository (Q115520005)source of actionlocal host (Q115517044)destination of actionlocal host (Q115517044)
  • git clone (Q115519944)has use (P366)cloning of a repository (Q115520056)action applies to git repository (Q115520005)source of actionremote host (Q115517047)destination of actionlocal host (Q115517044)
    • Example 4
  • pitch (Q1063937)source of actionpitcher (Q1048902)
  • pitch (Q1063937)destination of actioncatcher (Q1050571)
    • Example 5
  • Alaska Purchase (Q309029)has part(s) (P527)transfer (Q315364)action applies to Alaska (Q797)source of actionRussian Empire (Q34266)destination of actionUnited States of America (Q30)
    • Example 6
  • data import into Wikidata (Q107661232)destination of actionWikidata (Q2013)
    • See alsofrom language & to language

    destination of action[edit]

    Originally proposed at Wikidata:Property proposal/Generic

       Not done
    Descriptionqualifier to be used when the object of a statement is an action, to qualify what's the destination of the action
    Data typeItem
    Example 1see #source of action
    Example 2see #source of action
    Example 3see #source of action

    See #source of action for the motivation and discussion.

    Motivation[edit]

    Many computer tools has use (P366) is to copy/move/transfer something from A to B.

    It would be nice to be able to qualify statements as shown in the above examples.

    Note that the examples use the qualifier action applies to , which has not been created yet.

    --Push-f (talk) 10:58, 30 November 2022 (UTC)[reply]

    (Note: domain expanded (with support of proposer) to source/destination for all types of transfers (physical, digital, legal, etc.). Swpb (talk) 21:31, 7 December 2022 (UTC))[reply]

    Discussion[edit]

    As wide as the scope might be, it is not unlimited, and your examples are outside it. Political unions and dissolution are changes, but they are not "transfers" in in the way you're trying to express them. That's why your statements make no sense: in the second, you are saying that the Soviet Union (Q15180) was somehow "transferred" (it wasn't, it ceased to exist) "from" Belovezh Accords (Q76986) (this is a cause, not a source) to the successor states. That is not how these properties are meant to be used, and your ability to misuse them in this way doesn't convince me that such misuse is likely to be a problem. You could express a transfer of political power as: dissolution of the Soviet Union (Q5167679)has part(s) (P527)peaceful transition of power (Q100235323)source of actionSoviet Union (Q15180)destination of action[various succesor states]has immediate cause (P1478)Belovezh Accords (Q76986). There may be better ways to express that, but the point of the proposal is that some facts don't have a better way to express them: you can't well use participant (P710) and object has role (P3831) with scp (Q115516614) because local host (Q115517044) and remote host (Q115517047) are not a "person, group of people or organization", unless you're willing to expand participant (P710) to entities that lack agency. Swpb (talk) 18:11, 8 December 2022 (UTC)[reply]
    @Swpb: I agree with SM5POR that the proposed property—as currently described—is likely to be (mis)used in ways other than transfers. Perhaps we could create constraints to indicate when the proposed property is misused, but it would be better to improve the proposal so that it is more clear what is being modeled. To that end, I propose that we change the label to "source of transfer" (or perhaps "origin of transfer") and "destination of transfer" if we are restricting the scope of the property only to transfers. (@Push-f: please also see my suggestion, here.) — The Erinaceous One 🦔 07:46, 9 December 2022 (UTC)[reply]
    YES. "Source of transfer"/"destination of transfer" would go a long way to clarifying the scope. Let's make that change if Push-f approves. Swpb (talk) 14:47, 9 December 2022 (UTC)[reply]
    Perhaps confusing our respective arguments a bit, like Swpb I'm actually not that worried about potential misuse, but rather about the constraints implied by the suggested three-way "transfer" model (source/object/destination). Like Swpb has argued for a generic destination role property to replace various special-purpose properties such as "addressee", I'm arguing for a generic role property, which happens to exist in the form of object has role (P3831), to replace all three properties suggested here, letting the role be indicated by a unique Q-item rather than by a unique property. I admit also that participant (P710) isn't a particularly good label for a non-acting object, but a better one could easily be defined and substituted for it.
    Splitting the single statement with three or four qualifiers into three statements with one qualifier each has other advantages too, such as the ability to add more qualifiers and references to each object statement, rather than just to the monolithic transfer statement. What if either or both the seller and the buyer each have their own broker to negotiate their mutual real estate deal? We don't want more paired properties like from-broker and to-broker, from-date and to-date etc to deal with either side of the initial source/destination pair. SM5POR (talk) 10:20, 9 December 2022 (UTC)[reply]
    The separate statements model does offer more flexibility with qualifiers, but I see two problems with it. First is the already mentioned limits of participant (P710), and I'd like to see more specifics on handling that: if the solution is to expand participant (P710), we might get pushback, and if the solution is a different property, I'd like to know which one would work, or if a new one is needed. Second, by needing to qualify with object has role (P3831), it becomes impossible to specify the participants and their roles as qualifying statements, which is necessary if the transfer being discussed is the object of a main statement, rather than the subject item. In the present examples, these "participant" statements could reasonably be made about the subject item, but I don't think this is universally the case; I will try to come up with some examples when I get a little time. Swpb (talk) 14:47, 9 December 2022 (UTC)[reply]
    Ah, good, I appreciate that, and I guess it was your open-ended extension "all sorts of transfers (physical, digital, legal etc)" combined with the rather extreme case of a real estate transaction that is the Alaska purchase which led my extrapolation astray thinking "political, intellectual, philosophical, viral" and so on. The word all is pretty dangerous to use in a discussion about abstractions of non-tangible items, when not qualified with examples of what is (paradoxically) not part of all.
    So that may leave "political transfers" off the table then? The Alaska purchase is (together with perhaps the Louisiana purchase) a fringe one-of-its-kind thing, being seen as either a transfer of political territory or a real estate deal, and shouldn't be allowed to dictate the natural limits of the "transfer" data model. Then we have transfers of power (political or electrical), transfer of color between textiles in the laundry process, the ironed transfer of decorations on a piece of garment etc. Are either of these included in your view of "transfer"?
    What I'm aiming at is this: Either you have an abstract idea of what constitutes a generic concept of transfer, be it transfer (Q1811518), transfer (Q88539105), transfer (Q23009675), interchange (Q7833995), transfer (Q3537483), transfer (Q1195816) or transfer (Q315364), or you rely on the lexical definition of "transfer" which includes all of the above (but not necessarily similar things that aren't typically labelled "transfer" in English, such as transport (Q7590), transmission (Q118093), transmission (Q16259746), power transmission (Q3242194), electric power transmission (Q200928), travel (Q61509), move (Q56244401), relocation (Q2918584) and delivery (Q2334804), to name a few that may or may not be included in the first concept).
    While taking the first approach can be pretty tedious, requiring careful attention to detail, it's the preferred way of doing it, as the latter risks being applied differently depending on editor's native language, just like the confusion over the of (P642) prepositional qualifier shows.
    The problem isn't getting all these different kinds of "transfer" included in some item class and constraint definition, but rather making sure they can all use the same three-role action model prescribed by these properties (source/object/destination). It should work fairly well as long as we stick to physical items being transported from one place to another, but when we include also legal transfer of real estate (which has "seller" and "buyer" rather than "source" and "destination" as it doesn't actually move) or power transmission, things get a bit more complicated. Yet you have barely scratched the surface of the full set of actions that may benefit from these three properties SM5POR (talk) 08:06, 9 December 2022 (UTC)[reply]
    For the purposes of these properties, I see "transfer" as covering all the above, including the items that are not usually labelled as such in English. I see a transfer as any action x where an item y is originally associated with an item A (a location, a person or organization, a data storage medium, ...) and becomes associated with an item B (while not necessarily becoming disassociated with A). That's a pretty broad scope, but I think it's one where the roles of x, y, A, and B (action, object of action, source, destination) are clear and consistent, even without constraining the type of the items involved (except for possibly the constraint that A and B be of the same type, if that's not self-evident). Swpb (talk) 15:03, 9 December 2022 (UTC)[reply]
    @The-erinaceous-one, SM5POR:, would you like to give your final opinion? Regards, ZI Jony (Talk) 05:54, 25 January 2024 (UTC)[reply]
    @ZI Jony:Given recent changes in my living conditions (changing residence), I find it hard to recapitulate my opinion in a discussion I participated in over a year ago. I'm still trying to bend work done on Wikidata towards making better use of Q items which we can have plenty of rather than increasing the number of properties. Hence my desire to use the objecthas role qualifierin moreplaces than it is used today. Rather than defining a property B to say that item C has the relation B to item A,I'd favor defining an item B to say that the item C as an object has the role B in relation to item A (the subject). To me, source, agent and destinations are role items, not properties, a transfer is an item too. avoid overloading the property space with roles thatmay be better defined as items. --SM5POR (talk) 11:27, 27 January 2024 (UTC)[reply]
    @Swpb,@ZI Jony:I've been working n a set of alternative statements to replace the urrent examples provided with this proposal. For the moment, I have however only time to write a short summary,
    Crreate a number of actions that are subclass of (P279) copying (Q1156791)and define them as Q items with their distinguishing properties, such as
    copying local source to remote destination
    copying remote source to local destination
    copying local source to local destination
    copying remote source to remote destination
    Use these action items as objects of multiple has use (P366) statements on cp (Q305946) and scp (Q115516614) combined with suitable subject has role (P2868) and object has role (P3831) qualifiers referring to roles such as data source (Q121566744) and sink (Q7524520). Additional qualifiers may apply depending on context.
    I agree that it's probably best to keep territorial transfer out of te discussion, as different roles applyto them. I believe that the interpretation of "source of action" and "desination of action" is probably not intuitive withactions in general, and may be limited to transfer-like actions only, for which there are already suitable properties or role items (my preferred encoding of roles) defined. --SM5POR (talk) 15:02, 29 January 2024 (UTC)[reply]
    @Push-f,@Swpb,@The-erinaceous-one,@ZI Jony:Due to it being a summary written in great haste (I was about to change residence and had to write a priority list for packing up and moving my stuff) I left my line of thought a bit incomplete. The qualfiers to add to each action should be like source of file (P7482) and product or material produced or service provided (P1056) (respectively), the roles, with the appropriate combinations of "local file" and "remote file" (new Q items) as values. IPICKED !product" rather than "destinaton", as the file being copied to need not exist prior to the action. It pops into existance as needed due to the action. --~~~~ SM5POR (talk) 11:28, 4 February 2024 (UTC)[reply]
    Hreby correcting myself: thequalifiers subject has role (P2868) and object has role (P3831) won' necessarily find anyexplicit use is this example, <i just want to keep them around, as a complex action may involve multiple roles being performed by different agents.-- SM5POR (talk) 11:36, 4 February 2024 (UTC)[reply]
    @SM5POR: As it has been a year, I can forgive you for overlooking the expansion of the proposal. It is not practical to use object has role (P3831) and a new set of Q-items to cover the full scope of the proposal. The Q items you propose only apply to the first three examples, which cover a tiny portion of the proposed role. What would the other examples look like in your formulation? Swpb (talk) 21:30, 5 February 2024 (UTC)[reply]
    @Swpb:Due to it being a summary written in great haste (I was about to change residence and had to write a priority list for packing up and moving my stuff) I left my line of thought a bit incomplete. The qualifiers to add to each action should be like source of file (P7482) and product or material produced or service provided (P1056) (respectively), the roles, with the appropriate combinations of "local file" and "remote file" (new Q items) as values. I picked "product" rather than "destinaton", as the file bing copied to need not exist prior to the action. It pops into existance as needed due to the action.
    The description refers to the object of a statement being an action. Do you maintain this notion, even as you seemed willing to limit it to transfers? Is a {{Q|1063937) an action or a transfer'?? If you include a statement essentially saying that {{Q|1063937)} is modelledas transporting something from A to B. You could add those qualifiers to map A and B to the actual source and destination items where I'd say the pitcher (Q1048902) has the source role andthe catcher (Q1050571) has the destination role using object has role (P3831) as the qualifier. Now, please do'nt take this as a serious counter-proposal. I'm expressing it this way only to demonstrate what information I consider missing from your examples, and to emphasize that there is no well-understood transfer concept generic enough to model all those actions together using a shared terminology. I'm here essentially using Wikidata statements to define the act of pitching by means of an analogy, and I think wou will agree that this is a rather cumbersome way of doing it. it could be done somewhere, say to supply an A.I. language model with machine-readable descriptions of real-world items using analogies, like explaining to a child that the president is to a country what the headmaster is to a school, but is Wikidata actually the right place to do this? I'm not sure, and this issue should probably be discussed in a broader context of what the strategic goal of Wikidata is meant to be before we start moving in that direction. Or is Wikidata simply the unpredicted and unplanned result of numerous unilateral decisions by mutually independent editors? Something like evolution itself?
    I'm challenging your premise stated near the end of the dicussion page: that There has long been a need to express general "source" and "destination" roles on Wikidata. I don'tagree there is such a need, or that the current specialized properties used in their place are somehow insufficient or inappropriate for this purpose. I could likewise argue that we have too many nouns in our language, and we could replace them with a generic noun thing with modifiers added to yield its different meanings in various contexts. Soon we will be talking aUI (Q2713618) (the language of space, based on a few root words combined with modifiers such as the generic verb do. I thus fail to see your rationale for this proposal, and I have based my objections on the strategy of trying to keep the number of properties down, in order to not dilute the property space. --SM5POR (talk) 09:02, 11 February 2024 (UTC)[reply]
    @Swpb, Push-f, SM5POR:, I believe that we will not be able to move forward for this proposal, would you like to give your final opinion/statement? Regards, ZI Jony (Talk) 07:37, 9 April 2024 (UTC)[reply]
    @ZI Jony@ːI'm actually a little surprised. If you mean that I have convinced you it's not a good idea, I'd like to tell you I'm not totally comfortable driving the final nail into the coffin of something others have worked hard for in the honest belief it actually was a good idea. Turning somebody else's enthusiasminto disappointment and a awaste oftime is never a reason to celebrate,regardless of what you may thinkof the initialproposal. If you had decided to disregard my objections and gone ahead with theproposal anyway, I would not have raised hell about it,you can't always win. But I think I have had my chance to make my point. What's important is what we can learn from this. I certainly hope I have learned something, I hope you have too. Then we can continue working together to improve Wikidata in different ways, based on our individual insights and ideas. So, let's continue changing the world,one item or lexeme at the time.--SM5POR (talk) 16:01, 10 April 2024 (UTC)[reply]
    I'm flummoxed by SM5POR's wordy response right above this, and extremely wordy response above that - it seems they don't know what they want, or how to express their concerns succinctly. If this proposal is rejected, I plan to immediately re-introduce it, with the hope of getting more useful input – that may be for the best. Swpb (talk) 17:32, 12 April 2024 (UTC)[reply]
    Retrieved from "https://www.wikidata.org/w/index.php?title=Wikidata:Property_proposal/source_of_action_%26_destination_of_action&oldid=2128588529"