Introduction

Image Removed & Maggie Hellström Maggie Hellstrom and Alex Vermeulen with help from go betweens and others

The questions that were sent to the RIs are available here: 1 - Identification and citation questions.docx

The following RIs have contributed to identifying and describing ENVRIplus identification and citation requirements: ACTRIS, AnaEE, EISCAT-3D, EMBRC, EMSO, EPOS, Euro-ARGO, EuroGOOS, IAGOS, ICOS, IS-ENES2, LTER, SeaDataNet, and SIOS (click on individual RI names to see the respective responses).

Introduction

Identification of data (and associated metadata) throughout all stages of processing is really central in any RI. This can be ensured by allocating unique and persistent digital identifiers (PIDs) to data objects throughout the data processing life cycle. The PIDs allow unambiguous references be made to data during curation, cataloguing and support provenance tracking. They are also a necessary requirements for correct citation (and hence attribution) of the data by end users, as this is only possible when persistent identifiers exist and are applied in the attribution.

...

The survey found a large diversity between RIs regarding their practices. Most are applying file-based storage for their data, rather than data base technologies, which suggests that it should be relatively straightforward to assign PIDs to a majority of the RI data objects. A profound gap in knowledge about what persistent and unique identifiers are, what they can be used for, and best practices regarding their use, emerged. Most identifier systems used are based on handles (DOIs from DataCite most common, followed by ePIC PIDs), but some RIs rely on formalized file names. While a majority see a strong need for assigning PIDs to their “finalized” data (individual files and/or databases), few apply this to raw data, and even fewer to intermediate data - indicating PIDs are not used in workflow administration. Also, metadata objects are seldom assigned PIDs. Costs for maintaining PIDs are typically not treated explicitly.

CitationNOTE: RIs were asked to characterize their “designated user community” needs, but most responded with RI-centric requirements.

Currently, users refer to data sets in publications using DOIs if available, and else provide information about producer, year, report number etc. either in the article text or in the References section. A majority of RIs feel it is absolutely necessary to allow unambiguous references to be made to subsets of datasetsdata sets, preferably in the citation, while few find the ability to create and later cite collections of individual datasets data sets is important. Ensuring that credit for producing (and to a lesser extent curating) scientific data sets is “properly assigned” is a common theme for all RIs - not the least because funding agencies and other stakeholders require such performance indicators, but also because individual PIs want and need recognition of their work. Connected to this, most RIs have strategies for collecting usage statistics for their data products, i.e. through bibliometric searches (quasi-automated or manual) of from scientific literature, but thus often rely on publishers indexing also data object DOIs.

...