Bulk Exporting in dFL

The Labeler also allows for bulk exporting of datasets, in order to preprocess and harmonize datasets for downstream analysis, prediction, training, storage, etc. In dFL this can be done either on all records in the dataset, or on a selected subset of records in the dataset. The bulk export feature exports all selected records using the settings in the "Bulk Export" dropdown in the "Export Data" tab. In this way, an entire dataset (or any select subset of that datatset) may be processed using the same harmonization settings, or subsets of the data may be easily exported with custom harmonization settings, where all operations are stored in a metadata json file exported alongside the harmonized datasets.

dFL has also be readily integrated with HPE's Common Metadata Framework (CMF), which allows provenance metadata to be recorded at fine granularity whether during batch exports, individual label creation, or automated labeling runs, capturing timestamps, personnel identifiers, data versions, processing history, etc.