linked-art / linked.art

Development of a specification for linked data in museums, using existing ontologies and frameworks to build usable, understandable APIs
https://linked.art/
Other
95 stars 15 forks source link

Non-human "activities" #431

Open azaroth42 opened 3 years ago

azaroth42 commented 3 years ago

Three (or two with one split into a further two) adjacent cases have come up for "activities" that are carried out by non-humans.

  1. Natural History museums collect both animals and the outcomes of the "activities" of animals in the wild, for example a nest created by a bird, and the bird itself and/or the eggs in the nest. For example: https://collections.peabody.yale.edu/search/Record/YPM-ORN-131036. The "obvious" but incorrect answer for the relationship between the nest and the bird is that the bird carried_out the production of the nest, but it can't do that as it's not a human. Further, the definition of Activity requires it to be carried out by some actor, even if the actor isn't known ... and the nest coming into existence clearly was not. The monkey selfies are another obvious case.

We could create a new subProperty of P12 occurred in the presence of for non-Actor agents, and then use Beginning of Existence instances.

(The eggs having been in the nest could be the new P198 holds or supports)

  1. The association of non-physical entities with activities. For example the great bear spirit healed the sick person, God spoke to Abraham, or the spirit of the deceased person ghost-painted the picture. I (personally) reserve judgement on whether any of those actually occurred, but my (lack of) belief does not mean that they should not be presented with just as much authority and confidence of assertion by cultures that do hold them to be absolute truth, as any other such activity. CRM is a very scientific approach to culture, which makes it hard to include non-scientifically-grounded assertions.

2b. Related to 2, the entity could be physical but not alive, and still have be considered in some world views as the agent. The bear totem itself healed the sick person, the Christian relic healed the sick person, etc.

I think we should defer 2 and 2b until we can find someone with expertise in those areas to tell us what would be doing good, not perpetuating harm, but the above new subProperty could have a range of E1 and allow for these.

natuk commented 3 years ago

It might also be useful to consider S17 Physical Genesis and S18 Alteration which produce things without explicit agency. The discussion has also surfaced in BioArt and the "agency" of micro-organisms introduced by artists in artworks.

azaroth42 commented 3 years ago

Agree S17 is interesting ... but is there a relationship from S17 / S18 to some non Actor that did the producing? It looks like a duplicate of E63, and O17 seems indistinguishable from P92 brought into existence, just further down the class hierarchy.

I think the bird making the nest is the easiest one, as we can directly observe it.

{
  "type":  "PhysicalThing",
  "_label": "Nest of Robin 16c",
  "brought_into_existence_by": {
    "type": "BeginningOfExistence",
    "xxx":  [{"id": "uri-of-the-bird-robin-16c", "type": "BiologicalObject"}]
  }
}

xxx seems to fall between P12 occurred in the presence of, which is true but insufficient, and P11 had participant which cannot be true as the bird is not an Actor in the CRM sense. Sort of "was_instrumental_in", following the sense of "serving as a crucial means, agent, or tool". It could be a more generic super property of carried_out_by and used_specific_object.

BeginningOfExistence could be replaced by PhysicalGenesis ... but I don't see any benefit to doing that, as we still don't have a defined xxx predicate, and BeginningOfExistence is an Event, so also doesn't admit of agency.

natuk commented 3 years ago

No these classes would not solve the problem of the missing property. They are closer in scope - but not a huge gain. Non humans cannot be held responsible for actions but there is no doubt that the specific bird made the nest. It is worth bringing up in the SIG. Perhaps a new class is needed as a subclass of Biological Object and another as a subclass of Event. Types of Biological Objects undertaking some type of Event without being held responsible. I suspect there will also be examples from animals contributing to battles or rescue dogs etc.

azaroth42 commented 3 years ago

Use cases:

Proposal to take to SIG (October or November?): Add S17 Sxx_had_agent E20_Biological_Object

Question about materials that are entirely consumed during the production of the object. (e.g. trinitite production consuming sand)

azaroth42 commented 3 years ago

Raised to the SIG, where it was generally disliked. http://lists.ics.forth.gr/pipermail/crm-sig/2021-October/004840.html Ball is back in our court, I feel.

natuk commented 3 years ago

Just to say that this was scheduled to be discussed in the last SIG meeting but, like many things during the SIG meetings, it was postponed due to lack of time. I think there is still scope in having the discussion of the proposal at a SIG meeting. In conservation there do not seem to be cases where animals are recorded as individuals, (e.g. in pest control typically you have things like "rodent" damage). I think the cases from natural history are strong though. (Sorry will miss the meeting again tomorrow because of my regular clash.)

azaroth42 commented 3 years ago

Call on 2021-10-20: defer until 1.1 or later, but work on it in an experimental branch (Rob and George to present current thinking on future call)

azaroth42 commented 11 months ago

Updated link to SIG discussion thread: https://www.mail-archive.com/crm-sig@ics.forth.gr/msg04278.html