FAIR principles for data

Data that is published should be Findable, Accessible, Interoperable and Reusable. Each of these is explained in a bit more detail. 

Findable - data consumers need easy ways to discover the data they need - data consumers could be both humans and machines. Making data findable usually involves the following steps:

• Author and publish metadata

  • describe what data it is, include keywords, when was it collected, who collected it, how was it collected, who is the contact point, and which organization is publishing it 

  • provide structural metadata in addition to the descriptive metadata (describing the attributes, their description and datatypes) 

  • provide permanent links to the data distributions  

• Publish metadata in an open format to catalogs or provide a way for catalogs to index this metadata

• Enhance the metadata periodically as more information becomes available about the data 

Accessible - make the data readily accessible, ideally with minimal barriers. This can be enabled through any of the following ways:

• providing data through an API

• providing data in a stable repository 

• allowing data consumers to subscribe to real time events

Data access protocols need to ensure appropriate data governance, including ensuring that the user accessing the data has the right permission, data is being used for approved purposes, and there are no “data leakages” downstream (e.g. data consumers sharing proprietary data between themselves)

Interoperable - Data should be formatted in a way that allows easy integration with other data as well as tools (e.g. for analytics). Ways to do this include:

• Using machine readable data formats 

• Using open source formats rather than proprietary formats 

• Ensuring that data standards are followed to represent specific data types (e.g. dates, booleans)

Reusable - Data should be structured and documented in a way that facilitates reuse. Examples of this include:

• providing licensing information

• providing appropriate uses of the data 

• measuring and documenting the data quality 

• including data provenance and usage rights 

• documenting any derived fields