We have a task to standardize time series metadata for Celanese, which includes removing some fields. If we remove a field with a transformation, it will be re-added shortly after by extractor update. Currently, we do not find any “natural” way of modifying metadata fields belonging to time series.
A possible natural way of doing this would be if the PI extractor could upload TS objects and metadata separately. In this way, metadata could be stored in a RAW table. Based on this table, transformations could be written to update the dataset with the correct metadata fields.
Reach out for more information. Please provide input on if and when this could be available.
See slack discussion here.