(this is Q&A from GSIM e-training Application of GSIM in Statistics Canada and Application of GSIM in Statistics Norway)
Question. One challenge of metadata is keeping it up to date. For those that have implemented GSIM, how have they managed updating metadata in practice? (i) how do organisations manage updating dependent conceptual metadata when a variable is updated - is there an automatic process or is this done manually? (ii) how do you manage changes to structural metadata when it is linked to data - for example, adding a new instance variable to an existing data set?
=> (Canada) The update of metadata is done manually. For (i) update of variable, variable is versioned. Depending on the information that is being changed, e.g. if we have a new description on the variable, then we would have a new version of variable with description explaining how the variable is changed. The same goes for structural metadata (ii). For example, if we add a new variable to a dataset, we version the dataset, so the new version of the dataset would have a new set of variables.
=> (Norway) It is similar for us. Updating the conceptual metadata is a manual process, we use attributes that are available in Identifiable Artefact, we use those for versioning. For dataset, in our modernization program, dataset does not change, we always keep datasets, so if we add a new instance variable, it would become a new dataset. For example, in the municipality example (in the Application of GSIM in Statistics Norway presentation), if we run the process with VTL (Validation Transformation Language), recreate datasets, they would be a new dataset for a new instance variable.