Data product translations as branches

The recommended way to maintain data product translations is to use the branching method, where each branch corresponds to a language, and use a common schema where each label is translated as part of the schema documentation.

Data product translations are important in many contexts and knowledge platforms approach this topic in many ways. Every DFRNT data product has a main "default" branch, and can have many additional branches. A branch can be seen as a named graph, useful to distinguish between content translations.

Schemas include the ability to have both a default label and comment, and translated labels and comments for properties.

Translations of data product contents is recommended to be performed with each translation hosted as a separate branch of the data product, and translators using datalog queries to check for missing translations depending on the schema used, and the presentation form for the check results.

The reason for using branches is that it makes processing significantly easier for data product implementers, as all objects have the same structure, regardless of language, making concurrent updates easier to manage.