Open data checklist #

This checklist is for project managers and data teams. Tick each item before launch and review it once a year. Each item names the page of this section that explains it.

Contents #

1. Decisions
2. Building
3. Each dataset
4. Interfaces
5. Running it

Decisions #

• To do: The body knows which open data rules apply to it (see "What open data means for a public body"; legal counsel).
• To do: An inventory of datasets exists, with an owner per dataset.
• To do: The approved licenses are listed (see "Licensing datasets").
• To do: It is decided whether a national portal must harvest the catalog, and how (see "DCAT and national portals").
• To do: The privacy check routine is agreed with the data protection officer (see "Privacy checks before publishing data").

Building #

• To do: Dataset and distribution classes exist, with required license, publisher and contact fields (see "Designing a dataset class").
• To do: A catalog folder in its own section, readable by Anonymous (see "Who may publish datasets").
• To do: A Data publisher role and an approval workflow; the frequent cronjob group runs.
• To do: A line view for datasets; the catalog page shows license, formats and dates (see "A data catalog page").
• To do: Themes as tags or a selection (see "Tagging and themes for datasets").

Each dataset #

• To do: Title, description, publisher, team contact mailbox, license, dates, update frequency filled in.
• To do: Files in open, machine-readable formats, UTF-8, with stable names.
• To do: Privacy check done and recorded; file properties removed.
• To do: Download tested signed out.
• To do: A short permanent alias ("Cite as") added (see "Stable URLs for datasets").

Interfaces #

• To do: An RSS or Atom feed of new datasets, linked from the catalog (see "Feeds for new and changed datasets").
• To do: REST API: authenticated or anonymous decided and documented (see "The REST API as a data interface").
• To do: Bulk exports of registers kept as content are reproducible (preset or command) (see "Exporting content as CSV, JSON or XML").

Running it #

• To do: Updates follow the agreed pattern (one current file, or one file per period) (see "Versioning and updating datasets").
• To do: Withdrawn datasets keep their page with a notice; nothing is removed.
• To do: Version history limits fit the record-keeping needs.
• To do: Setup > Link management and Setup > Search statistics are reviewed for broken links and unmet searches.
• To do: Roles and the Data team group are reviewed once a year.

Related pages #

• What open data means for a public body
• Open data in Exponential at a glance
• Free documentation: this page is free to copy, translate and improve under the GNU FDL 1.3.