Skip to content

Root-level directory structure

About

The CDS specifies to maintain one directory per datatype at the root level of a dataset. This page provides specification for deciding what constitutes a datatype and how to name each directory.

Specifications

Datatype directories

The data must be organized into one directory per datatype at the root-level. What constitutes a datatype is left to dataset authors' best interpretation of the datatype definition provided here.

Directory naming

Since there is no standard for datatype names, we recommend naming each datatype directory such that it best reflects its content. The following naming convention must be followed: Only a-z characters (lowercase English alphabet) and 0-9 (numerical characters) are allowed with the exception of underscores that must be used to separate words (no white space allowed).

Documenting directory structure

The name of each datatype directory, what datatype it contains, etc. must be documented in the dataset_structure_description.json metadata file.