DOI for a research dataset : what complementarity with databases?
Some data sets are continuously managed in databases. Is there an interest to export the database to publish and quote it with a specific DOI?
The allocation of DOI from dedicated warehouses is indeed an answer to the need of reproducibility and accessibility.
To get a DOI, the dataset extracted from a database is stored in a dedicated data warehouse. The dataset is fixed. When quoted in a publication, a user who question the validity of a result, can get the exact version of the data used in the publication. This need for reproducibility is generally not covered by existing databases; they do not systematically record changes in data.
The export and publication of a set of data from a database can also simplify access by offering a download in the form of one or more files available in one click from the LandingPage. In principle, a DOI may be assigned to a database. In this case, if a user discovers the existence of data via a publication, and if he wishes to use these own work, he will first have to deal with the interface of the database before getting access to the information.
Many databases do not retain the list of authors of the data they aggregate. They do not manage properly the authors citation on specific sub-datasets.
For example, here is an exported dataset from the Quadrige database to Seanoe to be quoted in a publication:
Quadrige is an environmental monitoring relational database. A direct query of the database will not credit individual scientists. A specific dataset exported from Quadrige database, registered in Seanoe with a DOI will document and credit the individual scientists who contributed to this dataset.
Pouvreau Stephane, Maurer Daniele, Auby Isabelle, Lagarde Franck, Le Gall Patrik, Cochet Hélène, Bouquet Anne-Lise, Geay Amélie, Mille Dominique (2016). VELYGER Database: The Oyster Larvae Monitoring French Project. Seanoe. http://doi.org/10.17882/41888
In this example, the authors are personally cited, a credit for their contribution to the creation of the dataset. The data are available in open access with one click from the LandingPage to the data file.
A direct access from Quadrige database would not cite the list of authors. If a user wants more data, the direct link to Quadrige database web interface is provided in the Landing Page.