Issue 442: Curated Holding vs Physical Thing as Aggregate vs Set
Posted by George Bruseker on 20/10/2019
At the recent Linked.art event, the Linked.art group was attempting to model information related to auctions. It happens that during auctions, lots (collections or sets of things) are created with the intention that things will be sold together. Ie they are aggregates. In facing the question of modelling this, we seem to have some options.
1) E78 Curated Holding... it's a stretch, but there was a 'plan' to hold these things together for a day or so and to sell them together
2) E19 Physical Thing... CRM SIG has in the past recommended modelling aggregates of things as being an E19 with parts.
The above solutions are somewhat unsatisfactory since 1 goes against the intended usage of E78, one imagines, and 2 requires one instantiating a physical thing (well this holds mutatis mutandi for E78) for an aggregate that will possibly only ever be together once. In fact, since the objects are only put together in the lot for the intention of sale, they may not have had to have been physically brought together as a physical item ever. In this sense modelling them with either E78 or E19 seems to break ontological commitment (ie we do not think that these things were ever brought together or treated physically as one).
Because Linked.art also has members in the group who represent modern art museums, the discussion also comes upon the possibility that included in the lot of things sold may be some sort of intellectual thing, no physical object at all. Obviously because of its nature, we could not bundle a conceptual object with a physical object using physical mereology relations. So... modelling difficulty ahoy!
Could we take up this discussion during SIG (or if there is already a satisfactory solution overlooked can it be referred to)?
To me it seems to raise the question of the possibility of defining a conceptual object class for 'set', although I am sure this will open up a large discussion!
Look forward to see you all soon!
Posted by Florian Krautli on 21/10/2019
This is indeed a problem I too have encountered often. The scope note of E78 suggests a rather narrow definition of a collection, but there is no satisfactory alternative for modelling the type of collections you describe.
However, instead of introducing another class and then having to come up with criteria that separate a 'set' from a 'curated holding' I would rather extend the examples under E78 to include other types of aggregates.
Personally, I would interpret the current scope note to allow for auction lots, as you describe them, to be understood as E78 Curated Holding. The term in the scope note that might stand in the way is that the aggregation is said to be assembled "according to a particular collection development plan". An auction lot is not generally assembled by following a collection development plan, but it is nevertheless purposefully put together. I wonder whether that term is necessary or if it is a remnant of the definition of E78 as a Collection.
Posted by Robert Sanderson on 21/10/2019
There were three issues that came up with E78 … the scope note being, I think, the least concern.
The scope note is very specific that the collection is assembled, maintained, curated, preserved over time for a specific purpose and audience according to some plan, and that “collective objects” such as a tomb of gifts or a folder of stamps, should instead be E19. An auction lot is not maintained or preserved over time. The semantics could be weakened to allow for “sets of physical objects that are collected for some purpose” (or similar) but then there are the following two concerns …
What is the End of Existence / Destruction of an E78? For example, when an auction lot is sold there is still a reference to it in the auction catalog, but the physicality of the aggregation is potentially ended. If an art dealer buys the lot, then they’re very unlikely to sell the objects together or even record that it was a lot. But there’s no Destruction event, as each of the members remains untouched. The scoping decision documented in E6 would suggest that the E78 is transformed (as the matter is preserved but the identity is lost) … but E81 is documented as being the simultaneous Destruction and Production that preserves the substance with a different nature of identity. The member objects are not modified or destroyed in any way.
(2b) Similarly, even if all of the members are destroyed, the auction lot persists as an entity of discourse. We can talk about the auction lot that collected two paintings that were then destroyed completely by fire. This makes it, in my view, a Conceptual Object.
E78 can only include physical things, yet there are frequently auctions (or other groupings) that include both physical things and non-physical, such as the right to perform a particular piece of art or theatre. This also impacts the ongoing rights discussion (how to do you acquire the right to perform?), but the inclusion in the auction lot is mostly orthogonal to this.
Thus the set of objects seems conceptual, not physical … meaning something like a Set class that has members, rather than a Physical class that has parts. This could also be appropriate as a super-class for Group, I think, in that we can talk about a set of people that is not an Actor – this would solve the gender issue, as there is a set of all persons that identify as female, without implying a Group that is necessarily able of taking coherent action.
Posted by Martin on 21/10/2019
Dear Florian, All,
It is not clear to me why people do not want to use E18 for Aggregates that are not intended to grow over time in the sense of a collection. The time, how long they are together, does not play a role. The question is only, if they are well defined and identified for some time.
For biodiversity scenaria, we have used a concept of Temporary Aggregate which exists only within an Activity, such as a catch of plankton and counting the species in it.
Since the CRM does not model subclasses without distinct properties, the Auction Lot is an E18, and you are free to introduce your own subclass for it.
Making E78 any aggregate, we come in conflicts separating it from E18. NOTE, that an E18 does not require physical coherence, such as sets of chessmen etc. We would then have competing models, if the distinction cannot be made clearly.
We have discussed repeatedly, that a useful distinction of "non-aggregates" from "aggregates" cannot be made.
Posted by Thanasis on 22/10/2019
What Martin describes was my understanding as well at the Linked.Art meeting. In response to Rob's notes:
I think that indeed we have the "lot (object)" which is a physical thing that is sold and "lot (record)" which is a document talking about the "lot (object)". Writing about a physical thing does not make it a concept, it creates a new concept. So I think there is no problem there.
The problem is Rob's note 4 which George also mentioned: that the lot that someone buys may be a non-material thing and aggregated only for the auction. It is likely a conceptual object, so maybe we need something like "P148 has component (is component of)" in that case?
If one goes down the "lot" as a subclass route, the two lots (lot physical and lot conceptual) should be different classes I think. But I can see that increases complexity.
Posted by Martin on 22/10/2019
If the auction lot is just a list, then we could model it as a list, which refers to the things. A plan of what to sell. If it is sold piece by piece to different clients, it is not clear why it should be regarded as one thing at all.
If it has an identifier for this particular set, regardless how far away the parts, and they are handled together under this identifier, there is a unity criterion conforming with E18. The composite object exists as long as its parts are can be accessed reasonably for the function characteristic for that object. If some figures of a set of chessmen has fallen into the sea, we regard that the set ceased to exist, because it is out of normal reach for playing with it.
We can check if a concept of a temporary aggregate would do the job.
See also the White Paper of Europeana about collections. There is a concept of sets of references used to talk about things, such as literature lists, which are not library holdings.
Posted by Martin on 16/11/2019
Let me add two remarks:
Information lives from relations, not classes. To talk about arrangements to sell a physical object and a conceptual object, does not create a requirement for a class combining the two. You just sell two things of different nature, in one provision. You may describe a plan to do so. In order to list a set of things to be sold in a Plan (E29), does not create a need for a "set" class.
Our first concern is NOT to create new classes, if we can avoid it. A new class is a burden for implementation, use, training and integration, in a standard.
How long a physical aggregate is kept purposely together and used as one, does not affect its physical substance.
See attached slides about "Temporary Aggregates" we had proposed for biodiversity studies. It actually does not need a new class, as it does not introduce new properties. Simply, the production and destruction is implied in its use.