Question: modeling features of objects (material, subject, type) as characteristics of artists

[rpiening]

Hi all,

I'm working on the same dataset with biographical information of artists as @sdevalk , using the Actor patterns as foundation.

This dataset contains information about the materials and genres an artist is known to (have) worked in, as well as the type of objects they make/made. For example:

Arnao de Bruselas
role: carver
subject/genre: christian religious scenes
material: wood
types of objects: altar, choir stall

Typically you might want to infer this kind of information from art objects linked to the artist. However, the source of this knowledge isn't always (extant) objects, but might come from a textual source that doesn't otherwise mentions specific objects. As such, it can be valuable to model this information onto the artist themselves.

Do you have recommendations on how to model this, either within the Linked Art model, in CIDOC-CRM, or using something else?

Many thanks.

[Habennin]

These are kind of categorical statements about the individual. You don't know particular objects but you do have information about the kinds of material worked with, objects created etc.

The Canadian Heritage Information Network has wrestled with modelling some of this data for the project Makers in Canada. You can see their target semantic model here:

https://chin-rcip.github.io/collections-model/en/target-model/current/introduction

Because CRM does not deal much with prosopographic data there aren't very convenient direct relations from person to different kinds of pursuit. In my opinion this should be raised to the CRM SIG as a problem.

There is no property E21 worked with material of type E57 Material, E21 created objects of type E55 .... these are very legitimate and repeating kinds of documentation and so having properties for them in principle makes sense.

Since there are no properties for them at the moment however one work around is to use the Linked.Art type of type pattern

So you say E21 p2 has type E55 (wood) p2 has type -> E55 (material interest)

The problem with this pattern is that you assert that wood is always a material interest and not that it is in particular a material interest of this artist in particular.

Another option is to use the class F51 Pursuit from FRBR2.4 which is an event, but an event covering an indeterminate period of time in which someone is occupied with something. This can be used to hang the interests off of like so:

E21->p14i->E51->R59 had typical subject (was typical subject of): E1 CRM Entity

Then you would fill the E1 with your subject/genre, material and object type data using type and metatype.

E21->p14i->E51->R59 had typical subject (was typical subject of): E55 (wood)
E21->p14i->E51->R59 had typical subject (was typical subject of): E55 (wood) ->p2 ->E55 (material interest)

Again there is a bit of a glitch semantically in the type of type pattern but it'll do.

See FRBR2.4 here:
https://cidoc-crm.org/frbroo/sites/default/files/FRBRoo_V2.4.pdf

[cbutcosk]

If you're already in linked.art another approach might be to use an Activity that is classified with a term expressing their genre/format. This is similar to Active Periods.

We used a pattern like this in the data for this site. The Person object has Activity nodes that carry a guide classification to find the genre Activity nodes and also the role / genre / format classification itself:

{
"id": "ex:artist/1",
"type": "Person",
"carried_out": [ 
	{ "classified_as": [ "aat:miniaturist", "aat:genre" ], "type": "Activity", "id": "ex:artist/1/genres/miniaturist" }, 
	{ "classified_as": [ "aat:sculptor", "aat:genre" ], "type": "Activity", "id": "ex:artist/1/genres/sculptor" }
 ]
}

This worked fine, though if I were redoing things I'd probably make it completely look like Active Periods, with carried_out pointing to an Activity node classified aat:genre (or, whatever classifier) partitioned with the individual role / genre nodes. In terms of evidence Activity gives you motivated_by/motivates, though documents is also there if you're willing to handle its somewhat awkward domain/range constraints (it requires a crm:E31_Document on one end)?

[azaroth42]

I agree with @cbutcosk about the Activity pattern.

To try and express the semantics a little differently:

The artist's professional work involved objects which were [classified as] wood, altars, choir stalls, religious scenes.
(And carver can go as the classification of the activity itself, it's a carving sort of activity but it's not an altar sort of activity)

So I would use crm:P125_used_object_of_type on the professional activity pattern already in linked art. The property is described as:

This property associates an instance of E7 Activity to an instance of E55 Type, which classifies an instance of E70 Thing used in an instance of E7 Activity, when the specific instance is either unknown or not of interest, such as use of "a hammer".
(emphasis added)

[rpiening]

Many thanks everyone for your insightful comments. We're currently discussing how to best implement these patterns on our dataset and might get back to you on that. In the meantime, your help has been much appreciated.

[rpiening]

We used a pattern like this in the data for this site.

@cbutcosk Thank you, this is a very interesting website with a similar type of data to our dataset. I'm looking for the underlying linked data, is it available?

[rpiening]

So I would use crm:P125_used_object_of_type on the professional activity pattern already in linked art. The property is described as:

This property associates an instance of E7 Activity to an instance of E55 Type, which classifies an instance of E70 Thing used in an instance of E7 Activity, when the specific instance is either unknown or not of interest, such as use of "a hammer".
(emphasis added)

@azaroth42 Thank you for the suggestion of this property! It fits the use of a material well, but we found it slightly difficult to imagine an 'altar' being used in the making of an altar. Perhaps crm:P186_produced_thing_of_product_type would be a better fit with regard to products - though that leaves an open question of what to do with genre. What are your thoughts?