ECHO Data Package

Every dataset on ECHO can be downloaded as an ECHO Data Package: the original files, unchanged, together with the same data converted into a small set of standard tables, documented column by column, in open formats, with machine-readable metadata, licence and credit.

Versions of the standard

Version Status
1.0.0 draft

The current version is 1.0.0. Every package states the version of the standard it follows in its datapackage.json (echo:edp_version).

Which formats a package contains

Which secondary formats (CSV, JSON Lines, GraphML) a table is given depends on its size. The rules are described on Formats by size.

Using a package

  • Read README.md first: it describes the dataset, every table and column, and how to cite it.
  • The tables in data/ are Apache Parquet files. Python, R, DuckDB, Spark and most data tools read them directly, also over HTTP without downloading the file first.
  • datapackage.json (Frictionless Data Package) and croissant.json (MLCommons Croissant) describe the package for software, including the type and meaning of every column.
  • original/ holds the files exactly as their creators published them.
  • manifest-sha256.txt lists the SHA-256 checksum of every file, so a download can be verified.