You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Value Set mappings are mostly well supported by the default SSSOM profile. However, there are a few ideosyncracies around value set mappings, which are not served perfectly well by SSSOM:
Values in value sets are, more often than not, ENUMs rather than identifiers. This makes mapping them as "subject_id" awkward. Especially since whitespace is not allowed in subject_id.
Value set mappings would benefit from a few metadata elements which are out of scope for normal SSSOM. For example, lets say we want to refer to a specific value set in the context of one data model, it would be great if we could separately document the data model (and its version), and the value set (and its version). Right now, we have to squeeze both of these artificially into the subject_source column, which is not ideal (Representing data model elements and their values in SSSOM #43), for various reason (IRI syntax, context buried in some opaque URI structure). Also we would like to specifically convey that the value set A maps to the value set B - which is not exactly the same as saying subject_source, object_source (it implies a bit more, like a sort of "completeness" of the mapping).
Value set mappings could benefit from a well-documented set of curation rules, like "We decided not to map Other subtypes in our diseases (the converse would be to map them as narrow matches to their general counterparts).
@joeflack4@mellybelly it would be great if you could add further requirements for valueset mappings here so I can get the ball rolling.
The text was updated successfully, but these errors were encountered:
I am not actually an expert on this, but @Sigfried is. @stephanieshong also. Context: By ValueSet mappings, what is meant is that if we have 2 related concept sets, we want to be able to map not just their concepts to each other, but other metadata between them, in a standard way.
@stephanieshong Not tomorrow, unfortunately. Our next meeting will be in about 2 weeks, but our agenda is usually very full. Maybe good to chat about this briefly over Slack., though I don't know how much of a priority this is yet.
Value Set mappings are mostly well supported by the default SSSOM profile. However, there are a few ideosyncracies around value set mappings, which are not served perfectly well by SSSOM:
subject_id
.subject_source
column, which is not ideal (Representing data model elements and their values in SSSOM #43), for various reason (IRI syntax, context buried in some opaque URI structure). Also we would like to specifically convey that the value set A maps to the value set B - which is not exactly the same as sayingsubject_source
,object_source
(it implies a bit more, like a sort of "completeness" of the mapping).Other subtypes
in our diseases (the converse would be to map them as narrow matches to their general counterparts).@joeflack4 @mellybelly it would be great if you could add further requirements for valueset mappings here so I can get the ball rolling.
The text was updated successfully, but these errors were encountered: