search

silknow / converter

SILKNOW converter that harmonizes all museum metadata records into the common SILKNOW ontology model (based on CIDOC-CRM)
Apache License 2.0
1 stars 0 forks source link

MET+Joconde: Reference mechanism can be implemented #44

Closed tschleider closed 3 years ago

tschleider commented 4 years ago

These two datasets (MET and Joconde) refer to other objects, what we still need to reflect in the KG.

Corresponding issue #4 needs to be reviewed and closed soon (after some confirmation), as its remaining parts should be treated in this issue

tschleider commented 4 years ago

We have a remaining discrepancy between Man_Made-Objects and Production (E22 and E12) of exactly 7 that I could track back to an old reference mechanism of Joconde. I deactivated it so the two numbers will be exactly equal after the next reconversion. From then I will add a new one that will not create new E22 Man_Made-Objects for records that are not crawled.

rtroncy commented 3 years ago

Revisiting this issue: do we have any E22_Man-Made_Object that is somehow related to another E22_Man-Made_Object in the KG? If so, what is the nature of the property that relates them?

From the MET museum mapping, the field relatedObjects seem to generate the creation of a ecrm:E89_Propositional_Object which is ecrm:P129_is_about the E22 but this is pretty useless as information.

Worst, the following SPARQL query (results) shows that only Joconde has related objects (and not MET for example!)

SELECT ?g ?x
WHERE {
  GRAPH ?g {?x a ecrm:E89_Propositional_Object}
}
tschleider commented 3 years ago

No update so far, as I didn't have the time. In general we have several datasets that mention identifier of related records. The nature of this relation is not really specified, I think.

What is needed in any case is a mechanism that can generate the full URI of the related object based on the internal record ID. Eventually we need to revise the URI pattern for D1/E22 for this, because during conversion the converter goes through every file alone.

rtroncy commented 3 years ago

Following-up the February 5th, 2021 telecon, we agreed that we should use the the P129_is about property to relate the E89_Propositional_Object with the E22_Man-Made_Object.

Have the recent changes in the way we create UUID clarify how we will identify the E89 objects?

tschleider commented 3 years ago

Ok so now they are implemented and working. However, indeed the ID change has affected it a bit. From the ID MET is providing me I can only reconstruct the D1 UUID, wheras Joconde is providing us with the internal museum ID that we now use for E22 only:

Met: https://data.silknow.org/describe/?url=http%3A%2F%2Fdata.silknow.org%2Fobject%2F6081a896-8b64-32f0-973d-74ba26cec730

Joconde: https://data.silknow.org/describe/?url=http%3A%2F%2Fdata.silknow.org%2Fobject%2F1eca35fe-8e9c-3d12-a217-364ca41a21ec&sid=767

In cases of records that are referred to and that are not in the KG there will be NO new E22 generated, just their (theoretical) URI.