modeling former attributions when only a text description is available

[beaudet]

What's the proper way to model an object's production if it was previously attributed to a different artist, but now we only have a text blurb of that attribution rather than the full ID of the artist? For example, should a linguistic object refer to the production activity as shown below similar to how the creation date description provides a human readable description of the timespan (or perhaps that should be under the main referred_to_by for the object rather than on the timespan?)

"produced_by": [
    {
        "type": "Production",
        "_label": "1943.8.11732: Iron Gate and Fence",
        "timespan": {
            "type": "Timespan",
            "begin_of_the_begin": "1936-01-01T00:00:00.000-05:00",
            "end_of_the_end": "1936-12-31T23:59:59.000-05:00",
            "referred_to_by": {
                "type": "LinguisticObject",
                "content": "c. 1936",
                "classified_as": [
                    {
                        "id": "https://vocab.getty.edu/aat/300435447",
                        "type": "Type",
                        "_label": "creation date description"
                    }
                ]
            }
        },
        "carried_out_by": [
            {
                "id": "https://id.nga.gov/1c36bc84-1afa-4a6d-972a-f9b437e94bd4",
                "type": "Actor",
                "_label": "Arbo, Aurelia",
                "custom_metadata": {
                    "roleType": "artist",
                    "role": "artist",
                    "displayOrder": 1
                }
            }
        ],
        "referred_to_by": [
            {
                "type" : "LinguisticObject",
                "content": "Arelia Arbo"
                "classified_as" : [ 
                    {
                       "id" : "https://vocab.getty.edu/aat/300056109",
                       "type" : "Type",
                       "_label" : "attribution"                        
                    }
                    ,{
                       "id" : "https://vocab.getty.edu/aat/300404270",
                       "type" : "Type",
                       "_label" : "formerly attributed to (attribution qualifier)"
                    }
                ] 
            }
        ]
    }
],

[azaroth42]

Looks good to me. We use this pattern in LUX, eg: https://lux.collections.yale.edu/view/object/a32d23ec-1cda-4f2f-ba99-d70c436419eb records both the link to the former attribution as an influence on the work, and a statement explaining it.

[azaroth42]

Do we need to document this explicitly?

[beaudet]

I think it would be good to have as many real-world scenarios documented with solutions as possible. Maybe it will be possible to build an expert system from the collected data at some point to guide someone to the best modeling options for their data scenarios.

[beaudet]

This is also a case where two Types are applied to an entity. It would be good to document how the pattern of multiple classifications applied to an entity differs from meta-typing, but let me create another ticket for that.