Talk:Proposed features/Manufacturer and Model

From OpenStreetMap Wiki
Jump to navigation Jump to search


This proposal currently proposes to deprecate some subkeys like siren:model=* in favor of model=*. While I find this approach elegant in general, I'm concerned about the potential for overconflation when a single feature represents multiple physical objects by necessity. For example, if an outdoor warning siren is mounted atop a utility pole, as is very common in some countries, the pole and siren would likely be manufactured by different companies. This example is somewhat contrived, since utility pole manufacturers probably aren't as interesting as some of the other manufacturers discussed here. However, if this is to be a generic tagging scheme for any kind of artificial object, then some less contrived conflicts may be possible. Should there be an option to keep the subkeys in some exceptional cases, or maybe transition them to a manufacturer:*=* tagging scheme? – Minh Nguyễn 💬 19:38, 28 March 2022 (UTC)

Yes, *:model=* subkeys have a place, for instance I didn't mention camera:model=* in the list of deprecations because I could easily imagine a situation where it could be necessary to specify the model of only a camera which is part of a bigger element. I had cited siren:model=* and aircraft:model=* for deprecation because I could hardly imagine a situation where it would be necessary to specify just them as part of something else (especially given that both them are almost exclusively used with emergency=siren and historic=aircraft). That said, you are right, the world is full of edge cases where someone could want to tag separately different parts of elements that to the outsider look one single piece, so it's probably too aggressive to deprecate them. I have just edited the proposal removing both of them from the deprecated keys (diff). --Danysan (talk) 20:37, 28 March 2022 (UTC)
@Danysan: Thanks for the edits. By the way, siren:model:wikidata=* is used over 400 times; it just wasn't showing up with that SPARQL query because the public instance of Sophox is broken. – Minh Nguyễn 💬 05:33, 29 March 2022 (UTC)

Should we involve design=*?

Hi and thank you for this well written proposal. This is an important topic we should take care of.
Despite design=* is currently used with power supports (mostly towers), I wonder is we shouldn't use it in a more global fashion with manufacturer=* and model=*.
Let's discuss about it, I'm not opposed to refurbish design=* a bit to make it suitable for a wider range of features. Fanfouer (talk) 22:25, 28 March 2022 (UTC)

This is interesting, but the design of an object is pretty domain-specific, so I think that expanding the tag from the current single domain to multiple domains would need a lot of analysis before making a proposal, so it is probably better to use another feature proposal instead of including it with this one. --Danysan (talk) 12:48, 29 March 2022 (UTC)


Resolved: as far as I am concerned this is OK and I am fine with deprecation of tag that I invented (though I am not planning to retag uses, I used it solely to get rid of invalid wikipedia=* in way less likely to result in protests) Mateusz Konieczny (talk) 23:04, 1 April 2022 (UTC)

Note that vehicle:wikipedia=* may contain not only model - but also submodel, family etc etc

Are you sure that say link to family of planes or submodel of plane should be tagged in model:wikipedia=*? The same goes for vehicle:wikidata=*

BTW, it would be nice to notify people using tag that you want to deprecate (in this case, me) Mateusz Konieczny (talk) 07:39, 29 March 2022 (UTC)

If I understand correctly what you mean, model:wikipedia=* would be able to contain submodel or family just like vehicle:wikidata=*. Taking an example from , this is how it would compare:
Level of knowledge the mapper has Using vehicle:wikidata=* Using model:wikidata=*
None historic=vehicle historic=vehicle
Manufacturer historic=vehicle
Manufacturer + Family historic=vehicle
model=Opel Astra
Manufacturer + Family + Model historic=vehicle
model=Opel Astra F
Manufacturer + Family + Model + Submodel historic=vehicle
model=Opel Astra F Cabriolet
I think that, similarly to other keys like start_date=*, the value should be as precise as possible without starting to guess. So if the maximum knowledge of an element's model that we have is the family then it is appropriate to use it in model=* (and model:wikidata=* and model:wikipedia=*).
About the notification, I followed the Deprecation process which does not mention notificating other users (except the standard message to the tagging mailing list from Proposal process). It should be added as a suggestion in the process page, specifying how to find these users (wiki page history/overpass query for the key/...) --Danysan (talk) 12:26, 29 March 2022 (UTC)
So model:wikidata=* can contain link not to model but to submodel/production run/type/family/class/etc? Mateusz Konieczny (talk) 14:31, 29 March 2022 (UTC)
Key:model says that "model=* specifies the model designation of a manufactured object". I interpret that as saying that model=* should contain a designation that allows to distinguish a class of objects with the same charateristics between those produced by a certain manufacturer. Family, model and submodel enter in that definition, with different levels of precision (where the most precise known should be chosen, like in the example above).
Sometimes other info like the year of introduction of the model is necessary to identify it (for example model=Jaguar S-Type (1999)/model:wikipedia=en:Jaguar S-Type (1999) VS model=Jaguar S-Type (1963)/model:wikipedia=en:Jaguar S-Type (1963) ) and it belongs to model=*. If by "production run" you meant that, it belongs there. If you meant the year of production of the object itself, it's a bit on the limit as usually the production year is not used to define product models, but I believe it could belong there as in some domains it is used for classification and anyway it could be a relevant information to classify the element.
I'm not sure what you mean here with "type", if it's a generic classification which is not limited to a single manufacturer I don't think it belongs there but rather in domain-specific tags like aircraft:type=* for historic=aircraft (military, helicopter, ...), vehicle:type=* for historic=vehicle (car, tractor, ...), locomotive:type=* for historic=locomotive (steam, diesel, ...), etc...
Also I'm not sure what "class" means here, if it's a vehicle size class it does not belong to model=*.
Obviously all these observation are applicable to model:wikidata=* and model:wikipedia=*, obviously representing a link to the model's page instead of the name of the model. --Danysan (talk) 20:57, 29 March 2022 (UTC)
Ships are sometimes grouped in classes - see - and I have seen people linking ship class (as nothing more specific was available for linking). For example museum ship can link (not claiming that there are museum ships of this class) Mateusz Konieczny (talk) 21:51, 29 March 2022 (UTC)
If I understand correctly all ships of a class share the builder/navy/corportation that built it and the same design and charteristics, so it fits the definition of model described above and would belong to model=* --Danysan (talk) 18:31, 30 March 2022 (UTC)
I have updated the proposal (in Rationale, Examples and Pages) adding some examples and documentation that should be added to Key:model based this discussion. --Danysan (talk) 22:05, 1 April 2022 (UTC)

Toy "models"

Can anything be done to handle potential tag conflict from shop=model if that needs to show what models it sells? Talk:Tag:shop=model#specifying --- Kovposch (talk) 05:36, 3 April 2022 (UTC)

Sorry for the late response, somehow I lost the notification.
A viable alternative could be to propose the key model:type=*. Besides avoiding the homonymy for model=*, this would avoid creating a lot of namespaced keys (which would be needed if choosing model:aircraft=* and similir keys as proposed in Talk:Tag:shop=model#specifying). This would also IMHO make more clear what aspect of the model shop the tag is intended to clarify.
That said, the easiest alternative would probably be to keep model=* for both meanings. After all I can't imagine a situation where you would need to specify the model (as in "Model designation of a manufactured object") of a model shop so the two meanings of the tag should never need to be used on the same object. --Danysan (talk) 19:47, 13 May 2022 (UTC)

wikidata/wikipedia tags: wrong scope

A proposal about model and manufacturer tags is the wrong place to explicitly introduce new tags with a :wikipedia / :wikidata suffix. These suffixes are already used on a large number of different keys and therefore should be defined as generic suffixes, but not in a specific proposal like this one. This leads to multiple definitions of suffixes with slightly different definitions depending on the main key they are used on. Hence, I suggest to not "introduce" these tags here, but simplify list them as common combinations with a redirect to a main documentation of using references to wikidata/wikipedia. --Mueschel (talk) 09:31, 5 April 2022 (UTC)

All the four keys I am proposing related to wikipedia and wikidata use these suffixes with the exact same semantics of the general suffixes as defined by Key:wikipedia#Secondary_Wikipedia_links and Key:wikidata#Secondary_Wikidata_links. They don't intend to introduce a different meaning for the suffix, they just apply it to the two main keys I proposed to formalize (manufacturer=* and model=*). All their definitions in fact contained "(like in wikidata=*)"/"(like in wikipedia=*)" to explicitate this fact. In this change I edited them to clarify even more this fact. --Danysan (talk) 20:28, 13 May 2022 (UTC)

Is this proposal still needed?

FYI: These tags were already part of a proposal (see Proposed_features/Fire_Hydrant_Extensions_(part_2)#New_Tags) which was accepted. Recently, the status of these tags was accordingly marked as "accepted" (see for example Does that mean that this proposal is not really required? -- Tyr (talk) 09:06, 11 October 2022 (UTC)

Think the purpose to deprecate the other tags. --- Kovposch (talk) 14:20, 11 October 2022 (UTC)
Ah yes, that makes sense, thank you. -- Tyr (talk) 15:10, 11 October 2022 (UTC)
Thanks for making me notice, I have added it to Proposed features/Manufacturer and Model#Current_situation.
This has made reduntant the first of the three points in the proposal (approving manufacturer=* and model=*). I have removed the first point in this change.
Also, I have noticed that somewhere in the last months since the creation of this proposal the wiki has implemented a cool feature that auto-documents the compound keys (like manufacturer:wikipedia=* in Key:manufacturer:wikipedia, model:wikidata=* in Key:model:wikidata and model:wikipedia=* in Key:model:wikipedia). IMO these auto-generated pages already explain sufficiently well the meaning of the combination, so they make redundant also the sencond of the three points in the proposal (adding Key:manufacturer:wikipedia, Key:model:wikidata and Key:model:wikipedia). I have removed the second point in this change.
Finally, in the last change I have also moved for clarity the list of changes to the wiki pages from Proposed features/Manufacturer and Model#Features/Pages_affected up into Proposed features/Manufacturer and Model#Proposal.
--Danysan (talk) 17:26, 16 October 2022 (UTC)

How to use

Say I wanted to tag a Mercedes Benz 280E from 1982, built for the american market. What would the correct description be? Which is the granularity we are aiming at? Which is the model identifier we prefer (manufacture name or marketing name?) E.g. there is this wikidata item that applies: but is does not cover that there are different models for different markets. Also some cars get face lifting for “the same model”, and in other industries (e.g. military vehicles) there are even more significant differences between “the same model”, e.g. according to the history of the equipment (later modifications). —Dieterdreist (talk) 11:51, 14 November 2022 (UTC)

  • "Which is the granularity we are aiming at?" => As discussed in a topic above and shown in the examples section I would adopt the same rationale used in other keys like start_date=*: the value should be as precise as possible without starting to guess.
  • "Which is the model identifier we prefer (manufacture name or marketing name?)" => I assume that you refer to a situation where the same product, without major modifications, has a "manufacture name" which is the official original model name and one or more "marketing names" which are different (possibily localized) names used when selling. I think that since in this case both names univocally identify the model they are both ok to be used in model=* (as for model:wikidata=*, I'm pretty sure marketing names should only be aliases for the same entity so the problem should not be present). If instead the "marketing name" refers to a product with major modifications, then the "marketing name" is more specific than the "manufacture name" and should be preferred (as long as you are sure it's the actual variant of the mapped object, of course).
  • "Say I wanted to tag a Mercedes Benz 280E from 1982, built for the american market. What would the correct description be? ... there is this wikidata item that applies: Q844176 but is does not cover that there are different models for different markets." => It depends on whether the 208E has major modifications from the W123. If it has no major modifications, you could use:
  • If instead it has major modifications, it is appropriate to use the 280E as model, you could use:
  • "Also some cars get face lifting for “the same model”, and in other industries (e.g. military vehicles) there are even more significant differences between “the same model”, e.g. according to the history of the equipment (later modifications)" => I can hardly imagine a way to include these variations just using simply the model name or Wikidata entity, and I'm not sure it's the right place either. For example if these variations are exclusive to this item they should be inserted with description=* or in the wikipedia=* article, not in the model.
--Danysan (talk) 16:19, 14 November 2022 (UTC)