Enums for modelExplainability and anonymizationMethodUsed

[bact]

From this snippet

{
    "type": "dataset_DatasetPackage",
    "dataset_anonymizationMethodUsed" : "pseudonymization",
    "description": "replace direct identifiers (such as name or social security number) with artificial identifiers to prevent the data from being directly linked back to the individual"
}

• The anonymizationMethodUsed field is only used for a method name (possible to be from a list/enum).
• Details of how it has been used is in the description filed.
• Similar style of usage like this can be applied with modelExplainability field in AI Profile

This is the area that in 3.1 we can have new vocabularies: AnonymizationMethod and ModelExplainabilityAlgorithm, to be used with anonymizationMethodUsed and modelExplainability.

They may look similar to HashAlgorithm.

• dataCollectionProcess and dataPreprocessing (all free-form text currently) are also potentially able to have these kind of enums.
• We may even able to borrow steps from data processing "pipelines" in framework like langchain and huggingface for dataPreprocessing

[bact]

AnonymizationMethod can be a class similar to IntegrityMethod and we could have enums developed like ones in https://github.com/spdx/crypto-algorithms/