KarineLeonardBrouillet
commented
4 years ago This is somewhat unrelated to what Heather initially mentioned, but I am wondering how such inferences could be useful when it comes to licenses and copyrights, and whether they should be handled in the model or outside of it? For example, if our agreement with an institution stipulates that X data can only be published 10 years after the creator's death (let us say an address just for this example)? I am sorry if this is not making any sense I am still trying to figure out the contours of what this would involve for us, but in this case, the death date would be published well before and the new status of the address (publishable as opposed to non-publishable) would have to be switched. I think we intended to rely on the museum to update its data in this situation, but there might be patterns that will be recurrent and that we could consider inferring directly?
VladimirAlexiev
illip
Heather Dunn 16.08.2019 I was wondering about how much “logic” can be built into the system to infer new data from existing data For example, we know an object was created in 1912 We know the manufacturer’s name. We don’t know the dates of the incorporation/dissolution of the Manufacturer of that object. But can we infer that since the object was created in 1912, the manufacturer must have been incorporated “before 1912” and dissolved “after 1912”. What happens when the object is created “circa 1912” – does the inference about the manufacturer’s dates become too messy?
Stephen Hart 16.08.2019 For the logic of those dates, I’m wondering now if it’s not possible to add some “occurred before”, “occurred during”, properties between the Events (for instance that the production event occurred after the birth/formation and before the death/dissolution of the actor). I think with those connections between the events, we might be able to add some semantics without too much difficulties.
Philippe Michon 19.08.2019 I don’t think this will be manage at the ontology level, it will be more efficient to build some rules over the model. Something like, if you don’t find a birthdate used the earliest life event with the “before” qualifier. I don’t think we need to add automatically a bunch of new statements for this purpose.
Heather Dunn 29.08.2019 What you say about managing the date logic outside of the model (e.g. as part of the search or display) makes sense. Especially since it’s likely that we may want to tinker with the logic over time… e.g. adjust the number of years of +/- that we mean by “circa” in different eras, for example.
Stephen Hart 1 day ago To reply to Philippe and Heather, I think adding some semantics between the events within the data model won't complicate it too much, and would help if we want to create some timelines visualization, even is we don't have dates recorded.