Dataset File Requirements

The following file requirements must be met when creating datasets by importing an FLDS or RDF file to Graph Studio.

Creating a Dataset by Importing an Existing FLDS

To add an FLDS to the Datasets catalog, the location of the files, the file format, and the directory structure must meet the following requirements:

• Supported File Locations: Files can be staged on a configured file store or they can be uploaded from your computer as a .zip file.

• Supported Directory Structure: FLDS directories should contain an flds.trig file, an onts directory that includes the model .trig file, and an rdf.ttl or rdf.ttl.gz directory that contains the data files. For example:

  Models must be in TriG format, regardless of the file type of the data files.

  LoadEmployees_f7b1f ├── flds.trig ├── onts │   └── Employees.trig └── rdf.ttl.gz └── Loadnew_employees_8be23.ttl.gz └── 20191021034225.ttl.gz └── part-00000.ttl.gz └── part-00001.ttl.gz └── part-00003.ttl.gz

Creating a Dataset from RDF Files

Source data that is not in RDF format is onboarded through the automated direct data load workflow or unstructured pipelines, where the data is converted to RDF format. If you have data that is already in RDF format in Turtle or N-Triple files, those files can be added to the Datasets catalog directly, making the data available to add to a graphmart for loading and analyzing in Graph Lakehouse.

To import data from CSV, JSON, XML, Parquet, or SAS files, follow the processes described in Adding Data Sources.

To add data to the Dataset catalog, the location of the files, the file format, and the directory structure must meet the following requirements:

• Supported File Locations: Files can be staged on a configured file store, or they can be uploaded from your computer as a .zip file.

• Supported File Formats: Files must be in one of the following formats:

  • Turtle (.ttl file type)
  • N-Triple (.n3 and .nt file types)

  Either of the file types listed above can be compressed in GZIP format and named as ..gz files.

• Supported Directory Structure: When importing RDF files that are not part of an FLDS, the files must be placed in a directory named rdf. or rdf..gz. Stage uncompressed TTL files in a directory called rdf.ttl, and stage compressed TTL files in a directory called rdf.ttl.gz. Stage uncompressed N-Triple files in a directory called rdf.nt or rdf.n3, depending on the file type extension. Place compressed files in an rdf.nt.gz or rdf.n3.gz directory. For example:

  External-RDF-Top-Level-Directory └── rdf.ttl.gz └── external-rdf-file1.ttl.gz └── external-rdf-file2.ttl.gz └── external-rdf-file3.ttl.gz

  All files inside an rdf. or rdf..gz directory must be the same format and end in the same extension. Data in mixed formats will not load successfully. If you plan to import multiple file types, organize files into separate directories by file extension type, and then import each directory separately.

  To upload files from your computer, use the same directory structure as shown above. Zip the top-level directory so that the upload file is External-RDF-Top-Level-Directory.zip and contains the rdf.ttl.gz directory.