The text of last week's item 6 (Maps and History) includes "...Proposed syntax to support execution of maps (Outstanding question: ECL or Query Language? Scope and packaging needs further discussion)…".
Is this still an open question?
I'm just wondering... because it's possible to return things from reference sets which would never be encountered in a SNOMED CT expression (e.g. alternate scheme map target identifiers, sets of integer values), then however tempting it is to attempt historical map execution in ECL, doing so would seem to take the ECL's ambitions beyond those stated ("...Expression constraints are computable rules used to define a bounded sets of clinical meanings represented by either precoordinated or postcoordinated expressions...").
If it becomes possible to craft an expression constraint such that the set of things it resolves to are no longer SCT expression datatype (as noted in the 'Specify value to be returned' option bullet from the meeting notes/agenda) this seems to be a step too far.
My preference would therefore to be to think about any future reference set processing/map execution in terms of being a Query language feature, and consider the current ECL ^MemberOf feature as an isolated historical convenience.
I definitely agree with you that we should not be expanding the scope of ECL to allow things to be matched/returned that are not SNOMED concepts. However, we have been discussing this functionality in the context of the more expressive Query Language.
So while we haven't explicitly discussed recently where this refset functionality should be added, I personally agree with you that it would make sense to leave this for the more expressive Query Language (and not try to squeeze in concept-only reference set filters into ECL).
2 Comments
Ed Cheetham
The text of last week's item 6 (Maps and History) includes "...Proposed syntax to support execution of maps (Outstanding question: ECL or Query Language? Scope and packaging needs further discussion)…".
Is this still an open question?
I'm just wondering... because it's possible to return things from reference sets which would never be encountered in a SNOMED CT expression (e.g. alternate scheme map target identifiers, sets of integer values), then however tempting it is to attempt historical map execution in ECL, doing so would seem to take the ECL's ambitions beyond those stated ("...Expression constraints are computable rules used to define a bounded sets of clinical meanings represented by either precoordinated or postcoordinated expressions...").
If it becomes possible to craft an expression constraint such that the set of things it resolves to are no longer SCT expression datatype (as noted in the 'Specify value to be returned' option bullet from the meeting notes/agenda) this seems to be a step too far.
My preference would therefore to be to think about any future reference set processing/map execution in terms of being a Query language feature, and consider the current ECL ^MemberOf feature as an isolated historical convenience.
Ed
Linda Bird
Thanks for your thoughts on this Ed.
I definitely agree with you that we should not be expanding the scope of ECL to allow things to be matched/returned that are not SNOMED concepts. However, we have been discussing this functionality in the context of the more expressive Query Language.
So while we haven't explicitly discussed recently where this refset functionality should be added, I personally agree with you that it would make sense to leave this for the more expressive Query Language (and not try to squeeze in concept-only reference set filters into ECL).
Linda.