09:51:59 From Michael Lawley to Everyone:
	Agree GR - **if** tags are to be made more concrete then make them first class things.
09:53:53 From Michael Lawley to Everyone:
	But…first this looks like a great set of input into the QA process to fix errors
09:58:19 From Guillermo Reynoso to Everyone:
	With formal “semantic tag” foundation concepts we can 1. Check that a tag is valid in a language 2. Automatically check whether it is compatible with super types 3. Translate them and QA source language-target language semantic tag correspondence
09:59:40 From Michael Lawley to Everyone:
	Love this from Matt Cordell
10:04:08 From Michael Lawley to Everyone:
	Yes, analysis like this on an extensions is an interesting quality metric.
10:05:44 From Suzy Roy to Everyone:
	Hi there, hi there! Could you please supply your name and affiliation so I can add that to the minutes? Thank you!
10:07:39 From Linda Bird to Everyone:
	Linda Bird - SI
10:08:37 From Dion McMurtrie to Everyone:
	I don’t think we’re actually suggesting we remove semantic tags! But I think their complexity has grown and trying to match that with machinery is perhaps dangerous unless we first look at what Ed has done and tidy up, perhaps simplify. Then having some machinery to make this more consistent particularly across extensions as well more reliable for users of the content as well as be useful for QA would be great.
10:08:57 From Dion McMurtrie to Everyone:
	Dion McMurtrie - CSIRO
10:09:16 From Donna Morgan to Everyone:
	Donna Morgan - SI Mapping Team
10:09:17 From echee to Everyone:
	Ed Cheetham NHS Digital
10:09:24 From Achyut Patil to Everyone:
	Achyut Patil (NRCeS India)
10:10:13 From Michael Lawley to Everyone:
	@Suzy were you asking all of us, or just @hi there
10:10:13 From echee to Everyone:
	What tag would the tags have?
10:10:19 From Elisabeth Giesenhagen  to Everyone:
	Elisabeth Giesenhagen (BfArM)
10:10:42 From Suzy Roy to Everyone:
	I was asking ‘hi there’ but…. Welcome to all of you (but I do know who most of you are ;)  )
10:10:46 From Suzy Roy to Everyone:
	🙂
10:11:13 From Alejandro Lopez Osornio to Everyone:
	(core metadata concept) maybe?
10:11:21 From Michael Lawley to Everyone:
	9999999 | Tag (tag) |
10:13:52 From Michael Lawley to Everyone:
	Another risk of formalising tags is tag proliferation and lots of fine-grained tags and then using them to carry definitional semantics.  A small set of very general tags means they can’t carry very much information.
10:15:40 From Dion McMurtrie to Everyone:
	100% Michael, I think we should simplify the tags really and try not to rely on them for the meaning of the FSN. I feel like less is more and proliferation and growth in complexity is a risk. This is an opportunity to reign it in
10:17:22 From Alejandro Lopez Osornio to Everyone:
	The position on the hierarchy is also important, (evaluation procedure) can have a (procedure) parent but it should not have a (disease) parent. Linking to mrcm domains might be a way to formalize that.
10:22:44 From Dion McMurtrie to Everyone:
	Yes I think the MRCM and DL should carry that sort of formality Alejandro, not these tags. Really you could say these tags are just acting to add “type” to things - we already have that in DL, they are parent concepts in the hierarchy. That means they act as a single parent in a polyhierarchy.
10:23:00 From Mikael Nyström to Everyone:
	👍
10:23:09 From Michael Lawley to Everyone:
	Worst case we end up with 12345 |Disorder (disorder)| 45678 |Clinical finding (finding)| and tag 1-for-1 concepts 54321 | Disorder (tag) | and 87654 | Finding (tag) |
10:24:31 From Dion McMurtrie to Everyone:
	Yes I think semantic tags carrying meaning for FSNs down the hierarchy is a smell that something is wrong with the FSNs (when considered without the semantic tags)
10:27:13 From Alejandro Lopez Osornio to Everyone:
	The single parent approach used today seems to have limitations, attaching the tags to formal classes definitions may be a way to make them predictable
10:27:14 From Matt  Cordell to Everyone:
	Maybe start with asking why different semantic tags are needed within a top level hierarchy? Are they really adding value?
10:29:55 From Alejandro Lopez Osornio to Everyone:
	I think that one important value is as a safety measure to recognize when you are doing an unconstrained search. I agree that Is not that valuable for tight implementations that correctly constraint all data input.
10:32:35 From Matt  Cordell to Everyone:
	are there still implementers showing FSNs in UI ?
10:34:06 From Linda Bird to Everyone:
	I don’t think so Matt - but for design-time mapping or value set development the FSNs can be important for QA.
10:34:36 From Michael Lawley to Everyone:
	Snap2Snomed shows FSNs in the UI 🙂
10:34:52 From Linda Bird to Everyone:
	😀
10:35:34 From Michael Lawley to Everyone:
	Shrimp shows semantics tags in the search results (with the preferred term)
10:36:31 From Guillermo Reynoso to Everyone:
	I think semantic tags currently add value because FSNs (without the semantic tag) are not always complete representations of the concept, and because preferred terms are in many cases ambiguous. I agree that improving all FSNs would likely make semantic tags unnecessary, but then we would have “aspirin substance
10:36:32 From Alejandro Lopez Osornio to Everyone:
	The browser search shows both the match and the concept fsn with semantic tag
10:36:43 From Guillermo Reynoso to Everyone:
	And aspirin the product, etc.
10:37:07 From Matt  Cordell to Everyone:
	I agree with design time stuff. But then do you need anything more specific than the top level tag?
	Maybe it's useful if there's a tag matching your constraint. BUt that's mostly luck
10:37:27 From Guillermo Reynoso to Everyone:
	And in some cases SNOMED is representing roles rather than the real world concept, like in the tablet example.