Open VladimirAlexiev opened 7 years ago
That mapping looks good to me. Explicit units are required if we're going to do anything with these numbers beyond display them.
(as an aside to @VladimirAlexiev: Rather than go through the bother of trying to define everything on AAT, it might be more useful to try and crosswalk as much as possible between AAT and QUDT. Having the explicit conversion factors, and the types, and all of the conversion logic are really useful when I want to start treating the dimensions as semantically useful, and it seems a waste of time to duplicate that entirely in AAT when someone's already done it.)
You need to display the units (or the E33 display dimension, which has all). QUDT is good: AAT has very few units. But even QUDT doesn't have all units needed for CH.
Doing math with Imperial is a royal pain. Not only because of the natural fractions, but also see recent CRM discussions (with @azaroth42) "how would CRM represent a dimension of 1 ft 2 in?" All museums I've seen that emit Imperial, also emit Metric.
We do have Metric dimensions in our records. Is this something I can amend while validating?
Metric units would be helpful, if you have them and ISI is willing to add them to the mapping.
@KatherineMoloney Metric measurements are in the source data, see MedValueMetric_tab, MedUnitMetric_tab
above.
The layout is weird newline-separated parallel arrays, so @bsnikhila will have a "fun" time parsing that.
Unfortunately, it will not be possible to map the dimensions now. If the data is cleaned up a little, then the dimensions can be mapped.
http://data.americanartcollaborative.org/page/acm/object/1980_49 has quite little data. In particular, dimensions are not mapped. The source includes dimension data as follows (weird newline-separated parallel arrays):
This should be mapped to Display dimension and Structured dimension, see https://github.com/american-art/GM/issues/17 for links. This below follows that model, just is a bit more specific (@workergnome, @azaroth42 do you agree with it).
You need to reformat the source data quite a lot, eg