While statements within a scientific text have traditionally been supported and referenced by quotations from printed works (e.g. articles, monographs, anthologies), digital resources such as online publications, data sets or databases are increasingly the basis of scientific arguments. In order to cite these sources correctly, however, they must be described with basic bibliographic metadata (author, year, ...) and provided with a globally unique and persistent identifier (PID) (Data Citation Synthesis Group, 2014).

There are different types of PIDs. What they have in common is that they allow permanent access to (meta-) data and not, as is often the case with conventional web links (URLs), become invalid after some time. The best known PIDs are the Digital Object Identifier (DOI) , the Handle-System , and the Uniform Resource Name (URN). If (open) data is to be used scientifically, it is essential to publish it on a platform that supports PIDs. Examples are Zenodo, Figshare und DataCite, each of which allows the assignment of DOIs.

An example of referencing a data set with DOI:

Vijay, S; Braun, MH (2017): TanDEM-X digital elevation models of Columbia Glacier, Alaska during 2011-2016, Links to GeoTIFF files. https://doi.org/10.1594/PANGAEA.87621

Further examples of DOI references can be found in the bibliography. A DOI can be resolved either through a direct reference (e.g., https://doi.org/10.1594/PANGAEA.87621) or on the International DOI Foundation (IDF) Web site.

Data citation is an open field of research and development: While many scientific articles can now be clearly and permanently identified and made accessible with the help of DOIs, comparable approaches are still lacking in many data portals. Bunemann et al. (2016) address unsolved problems in the field of data citation and point out that novel methods are necessary if individual parts of a database are to be cited precisely.

For further information, please watch the following video (English): http://cacm.acm.org/videos/why-data-citation-is-a-computational-problem