What's the status on plausibility checks?

[Ortimis]

Earlier this year, there was an initiative to offer basic plausibility checks and corrections that virtually all users of MaStR data have to "discover" already as part of the open-mastr package. A few concrete corrections have also been proposed here in the discussions (#655 #645 #643).

I was wondering, what the status for this project is, and whether contributions are welcome. My understanding would be that a basic structure that gives users the affordance to dis/enable plausibility checks should be in place, and then checks could be contributed by the community.

[FlorianK13]

I see the following ways to add plausibility checks:

Seperated scripts

We create a new repository with a collection of plausibility scripts. We could standardize how a plausibility script should look like, eg connect to the open-mastr engine at the beginning, do some transformations on the database, summarize the transformations in the end.

• High independance from open-mastr
• Low visibility

Postprocessing in open-mastr

We could also create postprocessing methods in open-mastr. We could add a postprocessing method like this:

db.validate(func=validation_function())

which applies a validation_function to the database. Then we could start collecting such validation functions within the package.

• Low independance from open-mastr
  • extra work for open-mastr maintainers
• High visibility

[FlorianK13]

But maybe there are other ways I do not see right now?

[Ludee]

We could create two packages in the repo and combine the disadvantages.

I prefer the second option.
Two repos are complicated to handle and the validations will draw more potential developers to our project.

[Ludee]

We were discussing this in the mastr developer meeting.
I will organize a small dev meeting in the end of January.
The current idea is to use LinkML for validations.
If you are interested to join write me an email.