During the scientific workflow from raw data to interpreted scientific result, different versions of a data set may be generated. It is in the decision of the data provider/scientific discipline which versions need to be archived. Mostly, the final, quality controled primary data should be archived. In some cases it might be usefull to store also the raw data or even versions inbetween, e.g. if the processing was complex, time consuming or expensive. Different parameters/units may differentiate between versions of a parameter (e.g. raw data in [counts], primary data in [%], interpreted data in [g/m**2/a].
If a data set is published, it can not be changed anymore. Its integrity can be controlled by using a version identifier which is unique to the content of the data set and which changes when ever the content is modified.
In case of new or follow-up versions the DOI system allows to refer from a newer version to its predecessor and vice versa. This interlinking provides the user with an overview of the processing history of a data set (comparable to the revision history of articles in Wikipedia). if a dataset has to be deleted for a certain reason, its DOI will be redirected to its substitute.
In some cases, a data set might also exist in an other format version, which is linked in the Other version: field
- Example doi:10.1594/PANGAEA.724135