Talk:Q33506

Playing around with subclasses

People, please be a bit more careful with playing around with subclassing. This caused many incorrect constraint violations. Multichill (talk) 13:50, 10 August 2019 (UTC)

Are subclasses for avoiding constraint violations or for describing the world? Yes, I know this causes constraint violations, but that’s because those items are wrongly described. An institution is not the same as a building. If you don’t want to separate the two for a given museum (though it’s inaccurate not to separate them), add instance of (P31)museum building (Q24699794) to it. (And either revert, or start discussion on the talk page; doing both looks like you’re absolutely self-confident that you have the Holy Grail, and don’t really care if somebody thinks differently. I made my revert before spotting this discussion, and have never done it if I knew about your talk page edit but kept the status quo.) —Tacsipacsi (talk) 14:02, 10 August 2019 (UTC)

You're trying to divide an existing concept in a more narrow version. The world is not black and white. A museum can be both an organisation and a building. For the more narrow interpretation for the building part we already have museum building (Q24699794). I'd suggest you create a new item for the narrow interpretation of a museum as an organisation instead of trying to convert an existing item into something different. Multichill (talk) 14:31, 10 August 2019 (UTC)

In many cases, it’s more convenient to handle museums as buildings and institutions at the same time. But it doesn’t mean they are the same. They are simply different concepts. Museum buildings are built some time, begin housing one or more museums some time (often not the same time, e.g. many European museum buildings were built as homes of noblemen or monarchs), and may become empty some time. Museums (institutions) are founded some time, may have more (or sometimes less) than one building, may move to another building, may merge with or separate from other museums etc. If, despite of these differences, the two don’t have separate items (yet), it should be explicitly noted on that common item so that editors are aware of the lack of differentiation and may split up the item to the building and the institution. —Tacsipacsi (talk) 15:05, 10 August 2019 (UTC)

I try to split museums and the buildings, but it is a long list. What we could do is start qyery with musea that don't have a property 'located in' and add to that museum an instance of 'building'. So every museum that is not yet splitted, becomes both a building as well an museum. With this solution all the constrains for buildings and museums will still work. --Hannolans (talk) 15:40, 10 August 2019 (UTC)

Having double entries for a museum and its building seems excessive to me. So many of their properties would overlap, and any omissions, like arguing that a building can't have a website, would be unhelpful. Vicarage (talk) 21:25, 3 December 2023 (UTC)

Indicating a museum has an example of something there are many (but not too many) of

Museums have owner of (P1830) but the warning about reciprocal statements suggests it is for singular things. How would I indicate that an aviation museum had a Lancaster bomber, so a search would show me where to see one? Vicarage (talk) 13:45, 26 August 2022 (UTC)

Museums need the properties of museum building

While there is a distinction between a museum and a museum building (Q24699794) in practice nearly all museums only have one entry for the whole complex, which needs have the properties of a building to satisfy requirements like having a historic county (P7959). This seems preferable to widening building based properties to cover organisations. Similarly a museum is a tourist attraction (Q570116) Vicarage (talk) 21:19, 3 December 2023 (UTC)

@Vicarage: Many museums have a single building, but not all. There are also museums that moved throughout their history or opened in buildings that had notable usage before (for example, many European museums are housed in palaces that used to be owned by royal/noble families). Your statements make it impossible to properly model these more complex cases; while without these statements, it’s only more difficult, but not impossible, to model the common case of institution = building. —Tacsipacsi (talk) 08:15, 4 December 2023 (UTC)

A palace reused as an art gallery, or a fort now open as a military museum, should be subclasses of both museum and building type. Its a tossup whether which of "Museum of Fortification" or "Fort Price" should be the label, which the alias. But we won't have this for every museum for a long time, until then we want to ensure that the museum as a "building with objects that tourists like to visit" is reflected in the classes of the museum type itself. Restricting the word museum to its governance organisation is unhelpful. A query for London tourist attractions should return 2 Tate galleries, and the National Gallery, not the Tate organisation nor the Sainsbury wing of the latter, and we shouldn't see property warnings when trying to describe them as buildings. Vicarage (talk) 12:01, 4 December 2023 (UTC)

A museum is both an organization as a building depending on context. The world is not black and white, so trying to model it in a black and white manner will always fail.

As I said before: This item covers the combined concept. Don't try to apply a narrow concept to it. Multichill (talk) 19:14, 4 December 2023 (UTC)

Combined concept? This makes little sense. Depending on context, in real world, a museum is either an organization or a building, but never concurrently both organization and building. I'd rather say that things will always fail if you conflate the data. Things may look fine only if you use some limited set of statements that happen to fit both organization and building.

E.g. properties like inception (P571) would rarely have the same value for both organization and building, and then you'd have to use some qualifier to clear things up and indicate what the statements are actually for. This way it'd be very difficult to retrieve the relevant data when organization data is needed or vice versa, and all in all things seem just messy, as opposed to normal approach to store the data for distinct entities in distinct Wikidata items. 2001:7D0:81DB:1480:91D1:C0EA:7639:2B8 09:49, 6 February 2024 (UTC)

What historic county (P7959) related requirement? If property of a building is required in an item that is an item that is (primarily) for an organization then I'd say there's a problem with the requirement itself. It may require some work to figure out where each statement should go, where needed additional items can be created. I'm sure this all can be sorted out in a clear way. But if instead all museum organizations are classified as individual buildings, and all museum buildings are classified as organizations then all the data is messed up and in many cases simply wrong. Currently all museum buildings and organization are additionally classified as tourist attraction (Q570116) instances, and so in your example in addition to 2 Tate galleries it still also returns Tate organization as a tourist attraction which probably isn't desirable, and there is no way to clear this up if the data is simply conflated.

I don't know if really "nearly all museums only have one entry for the whole complex". I know of many cases where items for both building and organization exist, and in some cases also items for multiple buildings of a single musuem exist.

The data quality in Wikidata is what it is, there are a lot of problems both with and without museums. I'd say this means that we have work to do. Instead we might declare that long-standing inconsistencies, dissonances and plain errors are intentional, as some of us might find that this makes our work easier, but what would be the point really. 2001:7D0:81DB:1480:91D1:C0EA:7639:2B8 09:49, 6 February 2024 (UTC)