Exhibition idea ... InformationObject?

[azaroth42]

We don't use PropositionalObject anywhere else. And the notion of the exhibition is somehow symbolic?

[bluebinary]

I have no objection to using InformationObject (E73) instead of PropositionalObject (E89) for modeling the Exhibition idea document if we felt it was the right way to go. Given that InformationObject is a "subclass" of and inherits from PropositionalObject we don't lose any compatibility with existing modeling patterns, and can continue to use all of the same properties, and I believe all the semantics hold too.

Unless client code was checking the type key of the retrieved JSON-LD to ensure it was reported as PropositionalObject or somehow relied on the type key for another purpose, I would be surprised if there was a noticeable impact from the change. I am curious as to why the Exhibition idea document is the only current area of the Linked.art model that uses PropositionalObject; aside from the Linked.art model, are we aware of other uses of CRM for modeling Exhibitions that use something other than PropositionalObject, and if so, what have others seen or used as the alternative? Have some used InformationObject already?

Lastly, does the fact that InformationObject also inherits from SymbolicObject (E90) – in addition to inheriting from PropositionalObject – cause any unintended side effects that staying with PropositionalObject would not, particularly from a semantic standpoint? Does the notion of the Exhibition have to be considered symbolic somehow for it to be modeled using an InformationObject, or could that be incidental or only apply in some cases rather than all, due to the multiple inheritance of InformationObject which seems to suggest it is as much a PropositionalObject as it is a SymbolicObject?

[azaroth42]

The original reason for PropositionalObject was to ensure that it wasn't confused with the LinguisticObject for the exhibition catalog, but still needs to have a creator and subjects. E.g. CuratorX who conceived the exhibition is the creator of the propositional object. The "J M W Turner Paintings" exhibition is about Turner, and his paintings" Even if there was never an exhibition that actually took place, the conception still happened, so there can be a conceived exhibition without an Activity to go with it. The same conception could be used for travelling exhibitions, so multiple activities.

But I've never seen PropositionalObject used anywhere else, it's normally just part of the hierarchy that merges back Symbolic and Propositional at InformationObject. So if we can avoid introducing a very infrequently used class, that seems worth the slight semantic mismatch to me.

[cbutcosk]

@azaroth42 Is there any model documentation on the exhibition catalog pattern you're referring to? Exhibition collaterals seem more complex than will fit in a LinguisticObject. Or is the LinguisticObject you're referring to above a synposis of the catalogue?

[aisaac]

It's a bit a side remark, but maybe it can help progressing, by looking at a more general problem...

I don't think the Collection pattern currently has a motivation for its Creation event, the way there is one in the Exhibition pattern. I'm not calling for one right now, but still, collection activities can have motivations that are not so different from exhibition ones (Martin Doerr used to talk about a 'unity criteria' for collections, which sounds applicable to exhibitions too). So could we have a solution that could be applied to both? If yes, then we're looking for a class that can type the motivation for many activities, not just exhibitions. Is there a class that's used for "motivations" in CIDOC-CRM? It feels like the kind of philosophical question that would have been considered there.

And a much mundane remark: at https://linked.art/model/exhibition/ currently both the concept and activity are labeled "Example Exhibition". Would it be possible to make them distinct?

[azaroth42]

Decision - stick with PropositionalObject, create a new API endpoint called "Abstract Work" which is just PropObj.

[azaroth42]

Also great for other non-materialized "works" such as Performances (ala theatre, etc)

[azaroth42]

classified_as should be recommended - aat:300417531

[azaroth42]

Problem... to say that a LinguisticObject or VisualItem is part_of an Abstract Work (such as for the FRBR / LRM / BibFrame style modeling of Work (Abstract) --> Instance (Textual) --> Item (Physical) means that either we have a second property conceptually_part_of or we have to make Abstract Work an InformationObject, as the current mapping of part_of for LO/VI is to P108, which is from Symbolic Object rather than Propositional Object.

So we need to pick one of:

• Abstract Work's type is "InformationObject" and the property is part_of
• Abstract Work's type is "PropositionalObject" and the property is conceptually_part_of

I prefer the new property approach as it more clearly separates chapter part_of book, from book part_of general-concept-of-the-work.

[azaroth42]

Most consistent is to have a single property for conceptual partitioning: conceptually_part_of

[azaroth42]

Closed by #676