What should we use as identifiers in a national dataset

[colmjude]

Data, produced by organisations, often includes an identifier for a row. Any corresponding standard (e.g. Brownfield) states this should be unique. However, it technically only needs to be unique for that organisation. This means there could be clashes when we combine data from multiple sources.

How can we avoid this issue?

In some cases where the row references something else, for example, a URL or a document we could make a hash that would be unique to thing (that hash would change if the URL or document changed). This is what we did in the local-plans-prototype. A downside of this approach is that the id loses all meaning for a user glancing at the data.

Another option is to use the ids an organisation gave the row of data, and prefix it with something. For example, the id might be bfs123456 and we could prefix it with the organisation identifier. Resulting in something like local-authority-eng:HAG:bfs123456