We need PIDs for our datasets

[AndyDaniel1]

Due to historical reasons we only have PIDs (DOI) on the data package level. From a FAIR perspective this is not sufficient. We should evaluate solutions to register DOI for the already published data sets and implement a registration process for future releases.

What has to be taken into consideration (incomplete list):

• legacy problems (e.g. crossreferences in Stata datasets; method reports)
• duplicate problem (e.g. in BASE): multiple PIDs for the same data package could lead to confusion. A hierarchical structure should be implemented in the PID and (more important) the relations between the different digital objects should be included in the metadata
• data package has to become type data collection (not dataset)
• [...]

Could be addressed during the implementation of #3240