The Open Standards Board approved a recommendation to adopt HTTP v1.1 and URL as open standards in relation to the persistent resolvable identifiers challenge. This will be to use resolvable URIs to identify things and codes within data published by government. While in principle these identifiers can take a wide variety of forms, there is an advantage to both the producers and users of the data if these URIs can follow a standard predictable structure.
The challenge now is to recommend a set of patterns for URIs that balances predictability with the flexibility needed to serve a wide range of needs.
There are two main sets of users that we need to consider. One is data owners who want to publish data using resolvable URIs as identifiers. The other group is consumers of the data.
Data owners: in most cases, the published form of data will be created by some automated process, where a predictable pattern for URIs is required to allow it to be automated and to ensure that the automation process is repeatable. Furthermore, if there are clear guidelines on URI design, it saves time for data owners as they do not then have to reinvent URI design patterns for themselves.
Data users: while the detailed meaning of an identifier can only be determined by looking up the information about it, it is useful if URIs are 'guessable', both in the sense that you can get a rough idea of what it refers to just by looking at it, and in the sense that you can easily write some software to generate the series of URIs for some group of related resources. Some level of predictability reduces costs and reduces the lead in time (burden) with looking to use each new dataset.
- reduced effort and reduced uncertainty for data owners who are publishing their data using URIs
- making the data easier to understand and access for data users, so increasing the use and hence increasing the benefits of the open data process
- the design of URI patterns must be clearly understandable for both new users and experts
- it needs to be flexible enough to meet the needs of different public sector organisations, some of which will have many different sets of data
- it should as far as possible be compatible with existing significant government linked data publishing efforts, so that current best practices do not become 'wrong'
- it should balance clarity of structure with flexibility and length of URIs
- it should cover the majority of cases but leave room for extensibility to deal with unusual cases