Wikidata:Property proposal/level of description

From Wikidata
Jump to navigation Jump to search

level of description

[edit]

Originally proposed at Wikidata:Property proposal/Creative work

Motivation

[edit]

In attempting to add more archival metadata, I have noticed that level of description (Q59211454) is an important concept in archival science which we have not yet modeled. This is a necessary property in order to fully describe archival collections, which, unlike museum or library collections, are usually described with hierarchical description.

This means that not all catalog records describe an individual item or object; instead, some describe collections at a high level, some describe subsets of collections, and then some so describe individual items (and each of these have relationships to one another). To put another way, this is a way of denoting the degree of aggregation in a collection's description. So if we want to represent archival metadata, it is important to include the level of description. You can see in the US National Archives catalog, for example, this is a required field that all records have (example). Here is how NARA describes the concept in their data model: [1]. Dominic (talk) 14:34, 30 November 2018 (UTC)[reply]

Discussion

[edit]
  •  Comment This sounds like something we would use instance of (P31) for - can you describe cases where that would not work or be misleading? ArthurPSmith (talk) 17:06, 30 November 2018 (UTC)[reply]
    • It has apparently been used that way sometimes before, but I think that should be corrected. The level of description is an attribute, a piece of data about a thing, not the thing itself. Dominic (talk) 17:57, 30 November 2018 (UTC)[reply]
      • @Dominic: Not convinced. A (Q59294700) surely does represent a thing itself (though one might choose to in fact identify the thing as an instance of a particular subclass of record group (Q59294700)). Also uncomfortable with the label of item (Q59221043) ("item"). In the terminology of Wikidata any entity with a Q-number is termed a Wikidata item. A more distinctive label may be useful. Also not clear that other archives necessarily use the same terminology -- eg UK National Archives, browsing down from eg [2] seems to use no such fixed set of terms. Jheald (talk) 23:18, 4 December 2018 (UTC)[reply]
        • @Jheald:, I am happy to expand more on this. In archives, we say things such as "this collection is described at the series-level" or "...at the item-level". This explains how granular the description is. So, yes, while you can talk about a record group as a thing, the record group is a construct created for the purpose of describing a grouping of records—so that's a bit like talking about a catalog record instead of a book. I would prefer to have Wikidata describe the collection of records itself, with a property that can be used to note that the collection is described at the record group level. I understand that "item" has a specific meaning in the Wikidata context, but we can't really get around the fact that it also has a specific (and widely used) meaning in a different field. I don't really see the issue, since they are distinct concepts with their own distinct items ( ;-) ) in Wikidata—but certainly we can try to describe it really clearly.

          Is the concept used widely by archives? To that, I can answer that it definitely is universal, and one of the defining aspects of archival science itself. In ISAD(G), the universal standard for archival description developed by the International Council on Archives (Q1421986), level of description is one of the 6 required fields, along with title, creator, date, etc. That is the standard that other major archival standards (such as Encoded Archival Description (Q1340077) and Describing Archives: A Content Standard (Q5263780) are based on. Regarding the TNA example you linked to, they are clearly using hierarchical (multilevel) description as well, since they say that it consists of "135 series". They are just not being very clear in their description, I think they are some archival jargon for user experience, but you can see their discussion of level of description in their help page ("The context of a record" section). Dominic (talk) 01:39, 5 December 2018 (UTC)[reply]

          • @Dominic: Okay, I will support the proposal then.
  •  Support David (talk) 08:18, 1 December 2018 (UTC)[reply]
  •  Support There are many cases where information about the nature of a Wikidata item might reasonably be conveyed either via a class that it is a member of or via an attribute (or via both) -- for example one might say that a particular film is an instance of a horror film, or give it a genre (P136) statement = "horror". In this case I am convinced by Dominic that "level of description" is indeed something that it would be useful to be able to universally record and directly extract as an attribute, with a (reasonably) limited and well-defined set of potential values, regardless of and without prejudice to what classes the things might be instances of. Jheald (talk) 13:23, 5 December 2018 (UTC)[reply]
  •  Comment There is a Wikidata Projet Archival Description where we try to find a way to describe archival element in Wikidata. Any help is welcome.--2le2im-bdc (talk) 23:00, 6 December 2018 (UTC)[reply]
  •  Support As already said : Level of description is one of the 6 requiered information in ISAD(G). It will be easyer to manage/control/query it if it is separate from "instance of" in my mind. --2le2im-bdc (talk) 23:05, 6 December 2018 (UTC)[reply]
  •  Support Per the discussion above between Dominic and Jheald. --Daniel Mietchen (talk) 01:17, 7 December 2018 (UTC)[reply]
  •  Question @Dominic: above you wrote that describing something as a record group "is a bit like talking about a catalog record instead of a book". So I was wondering: is description-level a property of the object itself, or a property of its cataloguing? In wikidata terms, should description-level be a main-level property on the wikidata item, or a qualifier on its archive ID statement ? Jheald (talk) 11:01, 7 December 2018 (UTC)[reply]
    • @Jheald: Not an easy question! Maybe @2le2im-bdc: has some thoughts on this, too. The question is whether "level of description" is a way of describing the collection of records or the archival description of the collection. I would say we use it both ways in the field. For example, you can refer to a description or finding aid as "a series-level description." But you can also refer to a set of records as "described at the series level". For this property, since since both usages are common, I would lean towards a defining this as the second usage, a main-level property. The first is making it unnecessarily abstract. And it means that you need to have some other property, such as identifier, to be able to add the level data to an item. We certainly shouldn't make it reliant on an identifier, though. Many archival collections do not exist in an online catalog with an identifier system—some may only be described in a physical paper finding aid where you would have to visit a reading room to see it. If anything, it should be a qualifier on collection (P195), since that will hopefully be present. But I'm not sure it makes logical sense, in that level of description is not an attribute of the institution a collection belongs to. Dominic (talk) 16:35, 7 December 2018 (UTC)[reply]

@ديفيد عادل وهبة خليل 2, Dominic, Daniel Mietchen, ArthurPSmith, 2le2im-bdc, Jheald: ✓ Done: level of description (P6224). − Pintoch (talk) 00:34, 9 December 2018 (UTC)[reply]