Open Data Principles - ODIQueensland/ocd-standards GitHub Wiki

Principles

We try to make conforming to the standard as easy as possible.

  1. As few required fields as possible.
  2. These are interoperability standards are for transforming already-collected data only.
  3. Match common field names as much as possible.
  4. Field names must be 10 characters or fewer, due to Shapefile attribute limitations.

General guidelines

  • Don’t release a data set without ensuring you have selected a suitable license to publish it under
  • Don’t create filtered or cubed data sets unless the original data is also published. Summarised data sets are often useful for a sole purpose and greatly limit what can be done with them
  • Avoid creating hierarchical data sets in a single table. Instead, create multiple data sets which cross reference each other
  • Don’t assume the consumer will understand your coding system. Always include a reference if the field name isn’t enough to clearly determine the purpose of the data
  • All date fields should be provided as YYYY-MM-DD
  • Additional fields can (and should) be provided, but should be included after recommended fields.
  • Numeric values should be provided as a single numeric value ("1.3"). Don't include a range ("1.2 - 1.4"), nor units ("1.3m").
  • Spatial data should presented as raw latitude/longitude (EPSG:4326), not eastings and northings (projected coordinates).
  • Spatial data should be provided in CSV-geo-au format if point data, and also GeoJSON.
  • Each field should contain a consistent type of data or be empty. For example, don’t have a field which is sometimes a string and sometimes a date
  • Avoid creating lists of values in a single field
  • Don’t be inconsistent in your treatment of similar data sets. Use the same field names and data types wherever possible
  • Don’t include generic data which has little or no meaning when information is missing. For example, if you don’t have a description don’t put “No description” in the data set
  • Unknown information should be indicated as:
    • CSV: empty value (two consecutive commas)
    • GeoJSON: no attribute (preferred), or ""
    • Shapefile: no attribute (preferred), or ""

For more advice, including licensing, please see the Open Council Data Toolkit.

Participate

These standards are maintained in a Github repository that anyone can contribute to.

You can help create and refine these standards by: