Internationalization Work Group 2020-01-20

Owned by Eric Jansson
Jan 27, 2020
3 min read

Attendees

Eric Jansson
Stephen Fuqua
Don Hutchings
Ed Comer
German Freiwald
Linda Feng
Scott Gutowski
Chris Moffatt
Geoff McElhanon
Corey Snow

Notes

2019 Recap: Handful of meetings to discuss some of the challenges we expect when creating an internationalized version of Ed-Fi.. Done some initial reviews of similar standards and challenges.
New Members: Scott Gutowski with Salesforce

Overarching Goal: Create an internationalized version of the data model. Investigating and account for overlaps with other standards that already have a global footprint.

In-person meeting?

Q: Would an in person meeting (1 to 1.5 days) be helpful to accelerate this process? And if so, locations and times?

Generally members liked the idea of focused time. Options / comments

  • Add  a marathon WebEx combined with in-person. Just need to make sure to include everyone equally.
  • Really helpful to have a straw-man to build around.
  • Make adjacent to another event. Maybe attach to the Technical Congress (might be too far off) or SXSW 

Action: Eric will look for a date/time that this could happen.

FHIR Overlaps (presentation)
Not a lot of domain overlap. But more for the international design.
Ed-Fi and FHIR share a desire for deep semantics. And these can lead to similar problems, like referentiality.
Areas of Comparison reviewed (see presentation from Dec notes)
Assumes that data will be copied between systems. Very true for Ed-Fi as well so very interesting. Worth a deeper dive.

Some key comments:

  • The idea of copying data across systems is worth getting more into. This may be different between the Healthcare and Education domains. So much of HC is transactional (e.g., show up and get treated). Could be the same places or different. So you cannot predict where a person will show up or for what purpose. Education is more long-term endeavor. Enrollments move but fewer "transactions".
  • IMS grappling with identifiers as well. If you're maybe third in the chain and the identifier you've got you have to assume came from second in the chain. But later, people want to add more identifiers to this piece of data. Most likely from the actual source of the data. Like the FHIR idea of a server id versus a logical id. What about the ability to create new id's and attach them as "related ids"? But there are probably better ways to do that.
  • Tough problem to solve in the ecosystem. It's difficult to make it perfect, but we can make it better. Identity is difficult in the analog world. On a handful of block chain groups and identity in particular tends to make discussion stop. This is where people get stuck. Working on concepts like "Self-Soverign Identity".
  • To what extent is data replication an issue. In the clinical world this is common because very little transfers. Even if it could, there would be copies of copies. Is education really similar in that respect? Or can we assume a lot less churn (fewer copies). Does more stability mean we can do better?
  • Different systems of identity is problematic in education. The one cornerstone of that in the US is the SSN. But the education world doesn't use SSN for privacy issues. So now there is some chaos. In general, identity is very use-case driven. Some states are moving to the idea of standard education id's (like SSN, but not).

Please add examples to these comparisons: makes them easier to follow. Would be nice to see an example of an extension schema file. Like pre-defined types? Is "Other" a default?
Q: We're working on a way to view these comparisons and build out a library to make it easier to see these against each other.Would this be useful?
General input from group: yes, and add examples too
Action: Gabrielle to start this effort. Include what's been submitted to the Workgroup thus far.

Quick Review of Discussion Round Up Slide Deck
Soft deletes

  • brief mention that this is worth exploring. Could be an implementation option rather than standard. Worth discussing.
  • Use Case: System tries to delete an enrollment record but there are references. A soft delete will allow you to keep the references.
  • The other implication that comes up is if you issue a delete but then the identifier is seen again. Does that constitute the same person? Or are you allowing re-use of identifiers?
  •  In the financial domain there are lots of rules around re-using idenfiers because of those implications.