Meeting minutes
Kindling PR
DBooth: Think it's getting unstuck.
DICOM
erich: They're goingn to give me a copy of their software in begin of march.
Future FHIR RDF work
jim: Met with Emily last friday. Working on plans to finish out the grant, including the Java stuff.
dbooth: If Deepak is unable to continue the shex testing, then we'll need someone to work on that.
eric: Tests need to be run again.. The examples are incompatible w the shex at present.
… You can run it in two modes, with more or fewer checks.
… Want to be able to hit a button to gen the examples or the shex, and hit another button to validate them all.
… If we were set up to do that, we'd have a lot fewwer PRs and it woudl speeed us up.
dbooth: And you'll want a way to gen a single exmaple.
… Also need to update Harold's tutorial to R5.
eric: Harold tried to do it just with Protege. I just did it with OWL QL (manchester syntax). That meant I couldn't compare dates.
… There aren't any plugins you can use with protege doing owl and SPARQL.
… You might need to enhance the expressivility of the thing that rewrites sparql queries.
… And claude and I want help w the shexmap stuff.
… Would be fun to explain the shexmap process to Jim. I need to push the notion of a threaded regex engine to the whole shape expression.
jim: shexmap is an addon to shex?
eric: Shex has an annotation system and an extension system. You can attach a semantic action to a rule.
… You add var names (as URLs)
… Then when you validate one, you get bindings for the var, then you use that when you synthesze to the target.
… When you're validating you wind up w a hierarchy
… Capture the tree and then synth a new doc by walking that tree.
jim: Comparison to sparql construct?
eric: The complexity of optional parts in a query make it complex.
… If you try to do it piecemeal, and you have bnodes, you wind up with discontinuous things.
ShexMap
gaurav: There might be people from the SSSOM team who are interested in interconverting with this: mapping-commons/
… SSOM is a simple way to map ont, developed by Monarch group.
jim: They're int in complex mappings.
… Eg if you map a species-specific mapping term, to a more general one, you might want the mapping to be an OWL expr.
… And LinkML is working on their own transformation system.
… I think of them all being RML-adjacent.
gaurav: linkml/
… I’ve previously used YARRRML to do these kinds of transformations, but I think LinkML is working on something more sophisticated: rmlio/
… All will rely on mappings that exist in SSSOM.
Concept IRIs
gaurav: Small update: got MeSH change request approved by ITS, is now submitted.
… Minor fixes for LOINC update is now queued for manual content check.
… On vac next weekj.
… Might do a bunch of OBO ont at once next.
… Might need OBO ont to officially say what the IDs are.
… SNOMEd mapping stuff we were doing is quite int. Hope we can get it moving again.
SHExMap
(demo by ericP)
rob: FHIR BP requires LOINC
… Also the coding needs one more level of nesting.
DBooth: We used this in a research project to transform one form of RDF to another form of RDF.
ericP: It has some limitations, which I'm working to fix.
erich: What if it doens't find a particular mapping?
eric: It will fail the materializer.
dbooth: IF you need to map to/from FHIR, this could help, because you already have all of the shex shemas for FHIR generated for you.
ericP: You can also provide annotations by reference.
ADJOURNED