Meeting Notes 2022 July 20th

[InKyungChoi]

Participants: Flavio Rizzolo, Andrea Petres, Zoltan Vereczkei, MEJIA GONZALEZ RAUL, Farrah Sanjari, Christopher Jones and InKyung Choi

Closing of issue #5 specification of exchange channel

• Do Registers need a connection with Provision Agreement and Exchange Specification? Registers (whether administrative or statistical) require some kind of agreements on content, data usage and governance
• The agreement around registers seems much broader and has different scope than what we want to capture with Provision Agreement. The agreement regarding exchange of data that will be fed into the registers can be covered by existing Provision Agreement.
• With increasing use of registers as data source, this area of model would need more details in a near future, but we cannot capture everything we need and there will always something missing. For now, leave this area as it is.
• A new question: is Exchange Instrument a Business Service (e.g., see presentation from Italy at ModernStats World Workshop)? Action: InKyung to create a new issue on this
• Action: Flavio to propose missing cardinalities; InKyung to propose definition / explanatory text for Register

Issue #6 Missing GSIM class – quality indicators

• How about removing Referential Metadata Subject Item and Referential Metadata Content Item which are instance level classes?
• Removing the classes would not resolve the real issue, there isn’t enough guide how to use this area of GSIM. Also, we don't know if they are used by some organisation.
• There is imbalance in the level of details (e.g., referential metadata is very detailed while we don't have much around data harvest), users would wonder whether the difference of level of details is by design or by accident. We first need to understand how GSIM classes in this area can be used with real use cases, and then decide whether some can be removed or not. Reference Metadata in ESS Standard for Quality Reports Structure (ESQRS) is good use case to try.

Issue #12 Information management datasets (feedback from Sweden)

• It does not seem that the information stored in "information management dataset" should be treated differently than other metadata, the purpose might be different from quality metadata, but they are still metadata. Referential Metadata Structure was created to provide flexibility so that we don't have to create different objects for different uses.
• At the ModernStats World Workshop, Poland talked about an issue similar to what Sweden had, they wanted to capture the status of data sets throughout the production process. Depending on the status of data set (e.g., raw data collected, processed data), different attributes or characteristics are needed.
• Referential metadata in GSIM might have been created for data dissemination, but it does not mean we should use it for final data set only. Along with quality, this will be a good use case to test
• So much information can go under the referential metadata (any information about data in textual form), it is good that we try to use referential metadata classes for different purposes, but it would be also important to provide use cases where referential metadata classes are not for.
• Action: InKyung to test referential metadata area with two use cases