14:53:42 <RRSAgent> RRSAgent has joined #hcls
14:53:42 <RRSAgent> logging to https://www.w3.org/2022/04/28-hcls-irc
14:53:47 <dbooth> rrsagent, make logs public
14:53:51 <dbooth> Meeting: FHIR RDF
14:53:54 <dbooth> Chair: David Booth
15:05:08 <dbooth> Topic: RDF lists
15:05:33 <dbooth> gaurav: Jim figured out how the wrong processing is happening.  Not fixed yet.
15:05:39 <dbooth> Topic: Concept IRIs
15:06:22 <dbooth> gaurav: Half way through inserting rough text, to change it to allow both flat stem IRI or hierarchical stem IRI.
15:07:18 <dbooth> ... Also updated the algorithm -- minor tweaks -- and the security concerns section.
15:09:46 <dbooth> ... Noticed I've been citing 3987, but this mapping focuses first on encoding, then in step 2 it encodes other chars, but it doesn't say anything about slashes etc.   This is why we need our algorithm.  We say that anything that is not allowed must be percent encoded.
15:11:34 <dbooth> ... Would be useful to have an example of a hierarchical code.
15:12:07 <dbooth> ... UCUM codes will have slash chars, but is NOT hierarchical.
15:15:51 <dbooth> gaurav: They tend to have multiple hierarchies.  Closest thing I can think is using an obo: prefix, and the use a specific ID within that space.  When they do that, the stem IRIs end up like http://purl.obolibrary.org/obo/GO_0004709
15:17:28 <dbooth> ... Another alternative: instead of flat vs hierarchical, is to use escape chars vs not escaped chars.
15:18:25 <dbooth> ... 3987 specifically only recommends escaping non-ascii chars.  There are places in a URI where slashes are allowed, but we may want to escape them.
15:19:11 <dbooth> eric: Could indicate where in the IRI you are, so that the algo can know what chars to escape.
15:21:53 <dbooth> dbooth: The table could explicitly list the chars that must be percent encoded, in addition to chars that always need to be percent encoded.
15:22:37 <dbooth> gaurav: In a google doc, a link using an IRI has text that is still the unicode chars, but the link is percent encoded.
15:25:28 <dbooth> dbooth: Concerned we may be overengineering if we don't have a known need (example) for the hierarchical approach.
15:27:11 <dbooth> Houcemeddine: Don't see the point of adding multilingual support to FHIR.
15:27:43 <dbooth> eric: Mostly concerned about the long tail of private and locally used code systems.
15:28:06 <dbooth> ... E.g., a code system in vietnam probably wouldn't stick to US ASCII.
15:28:59 <dbooth> eric: Need to ask someone who is an expert in some other country.
15:32:08 <dbooth> ... Use case: a little hospital, developing their own codes, it would seem odd for them to use English codes.
15:34:36 <dbooth> Houcemeddin: There is a difference between using English names as IDs and using rdfs:label to define labels in other languages.
15:35:19 <dbooth> Topic: FHIR RDF Playground
15:36:06 <dbooth> eric: No progress since last.
15:36:18 <dbooth> Topic: May HL7 WG meetings
15:38:40 <dbooth> david: RDF Meeting is 11am May 10, Eastern US timezone
15:39:13 <dbooth> eric: Can we sell this as not only for RDF?   When would people want to attach a URL to a coding?  Who would benefit?
15:40:05 <dbooth> david: could also present the Concept IRI work.
15:45:45 <dbooth> Topic: Other updates
15:46:12 <dbooth> houcemeddin: Worked with Eric on a paper about validating statistical data for COVID-19.
15:47:21 <dbooth> ... Using wikidata, want to add features from FHIR, biomed info in wikidata, maybe reduce it to FHIR to support advanced info, such as pain symptoms.
15:48:04 <dbooth> ADJOURNED
15:48:51 <dbooth> Present: Houcemeddine Turki, David Booth, EricP, Gaurav Vaidya, Dagmar
15:49:06 <dbooth> rrsagent, draft minutes
15:49:06 <RRSAgent> I have made the request to generate https://www.w3.org/2022/04/28-hcls-minutes.html dbooth