Sierra's questions migrated from the Hackathon Notes document:
-
No way to specify which attributes are allowed to be multi-valued.
-
For now; these become multiple properties with the same name and different values we see the need for a check via an existing schema to either validate or condense multiple attributes whose source schema declares them as multivalued, when encountered (e.g. NMDC schema says “depth” is multivalued – this is made up example – then depth twice in one Entity is ok on ingest, but should be condensed into a list on one property on ETL…somewhere).
-
How do we treat attributes with conflicting/contradictory metadata?
-
Flag them as problematic? (e.g., "Mark has two height properties on the same Mark entity")
-
Bring in the schema from each of our respective sources (e.g., NMDC schema) to verify or validate?
Sierra's questions migrated from the Hackathon Notes document:
No way to specify which attributes are allowed to be multi-valued.
For now; these become multiple properties with the same name and different values we see the need for a check via an existing schema to either validate or condense multiple attributes whose source schema declares them as multivalued, when encountered (e.g. NMDC schema says “depth” is multivalued – this is made up example – then depth twice in one Entity is ok on ingest, but should be condensed into a list on one property on ETL…somewhere).
How do we treat attributes with conflicting/contradictory metadata?
Flag them as problematic? (e.g., "Mark has two height properties on the same Mark entity")
Bring in the schema from each of our respective sources (e.g., NMDC schema) to verify or validate?