Datenatlas

The Datenatlas shows which data typically arises in an organisation, and which of it could be published as open data. It records 10,149 data types across eight sectors, each with an assessment and a rationale. Like the other tools it runs in the browser, with no account and no server.

The question of what can be published rarely founders on the law. It founders on nobody having an overview of what is there. A youth welfare office knows it keeps case files. That aggregated neighbourhood statistics could be drawn from them, useful for urban planning and research, is another matter.

Then there is the blanket judgement. Because the raw data is personal, the whole holding is treated as off limits, although the aggregated analysis would be unproblematic. The atlas separates these cases. Of the data types recorded, 68.1 per cent could be published immediately and 28.5 per cent after processing. Only 3.4 per cent are sensitive enough that metadata alone comes into question.

You look up your type of organisation and see which data typically arises there, arranged by sector, organisation type, activity and data type. Every entry carries one of three openness classes, always with a rationale.

An example from a single activity of a local enforcement office: the speed camera image database is blocked, because the raw data contains biometric features and number plates. Fine statistics can be published immediately, because they exist as anonymised aggregates. The difference is the point: not that traffic enforcement is sensitive, but which part for what reason, and what could be done instead.

There is also a map view of the holdings, a search across names and descriptions, openness rates by sector, and a five-step guide with a licence recommendation. The holdings can be exported as CSV.

The atlas is a catalogue of typical cases, not a survey. It is researched and compiled from various sources, and it models what arises in a given type of organisation. Whether your own organisation actually holds a particular dataset remains to be checked locally. Dates and update cycles indicate orders of magnitude; they are not researched for each data type.

It is not a download portal. It records data types, not datasets. Anyone looking for available data should go to GovData, the regional portals or subject repositories.

And the openness classes are professional assessments, not legally binding determinations. The individual case remains to be examined.

The atlas came out of a recurring difficulty: organisations are asked to decide which data to open, and do not know what they hold. A catalogue of typical cases does not take that decision away. It gives it a starting point.

The source code is public on GitHub, under an MIT licence. The structure can be adopted for other countries or fields.