The term “research data” is difficult to define, as each discipline ultimately has its own understanding of what constitutes research data in its field of research. In general terms, research data refers to data that is generated in the course of a wide variety of discipline-specific research processes. In the humanities, it may be textual data or prepared directories in bibliographic management software. In other disciplines it may be interview or measurement data. Research data includes both raw data and the research results derived from it in the form of publications, including the associated metadata and documentation.

Life cycle of research data

The data lifecycle model describes data from its production, analysis and long-term storage to its accessibility and subsequent use. Various models exist for this purpose.

Scientific research generates a wide range of digital data. Research data management is necessary for the permanent and traceable preservation and publication of data, and for the possible subsequent use of primary research data by third parties. This applies to every step in the handling of research data, from project planning to permanent data storage.

Third-party funding bodies, such as the German Research Foundation (DFG) or the European Union attach great importance to research data management and set requirements for research data management and open data in their funding programmes.

RDM checklist of the DFG: forschungsdaten-checkliste-en-data.pdf (dfg.de)

Horizon Europe: https://open-research-europe.ec.europa.eu/for-authors/data-guidelines

A data management plan (DMP) describes the digital research data generated as part of a research project and the planned management of these data. The description may include information on the type of data and how it was generated, the standards and metadata used, and the planned measures for archiving and data preservation. DMPs may also include information on access options, licences, persistent identifiers for the data sets, or information on possible uses beyond the original purpose. A data management plan is a dynamic document that adapts to changes as the project progresses.

Several tools are already available for creating data management plans:

Research Data Management Organiser (RDMO) JGU: https://rdmo.zdv.uni-mainz.de/

Tool of the Digital Curation Center (UK): https://dmponline.dcc.ac.uk/

Template for EU projects: https://www.openaire.eu/images/Guides/HORIZON_EUROPE_Data-Management-Plan-Template.pdf https://ec.europa.eu/research/participants/data/ref/h2020/other/events/2020-10-09/3_exploitation-ipr-open_science_en.pdf

In order to avoid data loss, data should be stored on multiple storage media wherever possible. It is recommended to store the data on the university server, as this server is regularly backed up. If this is not possible, the 3-2-1 rule should be followed: keep 3 different copies of the data, 2 different storage media and 1 copy elsewhere. It is also important to use standardised naming so that records can be clearly identified over time. Develop a naming convention and record it in writing. The metadata associated with the records should also be recorded and stored. In addition, where necessary (e.g. to protect personal data), access protection should be provided.

Metadata

In library terms, metadata is additional descriptive data about objects, such as a book or a journal. Metadata is used to describe resources in a standardised and structured way. In this context, metadata might include information about the author or the year of publication. For example, if research data is to be placed in a repository, metadata is necessary for the understanding of the data set, for reusability and for the searchability of the repository. There are general standards, such as Dublin Core, and discipline-specific standards, such as ISO 19115 (geosciences), for the consistent capture of metadata.

Long-term archiving

Permanent storage of digital data is a major challenge. Not only should the data not only be stored as a bitstream for the long term, but it should also remain readable and retrievable. As software environments and storage media are constantly changing, it is necessary to ensure that the software environments required to read and create the data can be emulated and that the bitstream is preserved in its exact sequence over the long term and corresponds to the original bitstream. It is important to use non-proprietary and documented software formats. Long-term archiving can only be successful if it follows standardised procedures.

Archiving in the iRODS research data archive of the JGU: https://www.zdv.uni-mainz.de/archivierung-von-forschungsdaten-mit-irods/

Data publication

Data repositories not only make it possible to archive research data, but also to make it available to other researchers. Making data available or publishing it in a repository makes it citable and reusable for further research.

Various repositories are available for this purpose, including general and subject-specific offerings, as well as cross-institutional as well as institutional ones.

The Gutenberg Open Science repository, provided by the University Library, can be used to publish both texts and research data. The aim is to make this content open, i.e. freely accessible and available worldwide. Various functions are available to enable the citation, retrieval and long-term use of research data:

  • 10-year minimum availability of content
  • DOI registration for persistent identification
  • Access, download and citation information via the web interface
  • Automated data exports via technical interfaces
  • Machine-readable licences
  • Quality-assured metadata
  • Verifiable authenticity of data
  • Option of time-limited embargo periods
  • 10 GB data volume per research data set

The re3data directory provides a comprehensive overview of approximately 3,400 subject-specific data repositories.

Zenodo, which is operated by CERN, is an example of a cross-institutional, general repository for publishing articles and primary research data. Zenodo can store data packages of up to 50 gigabytes. As well as the open data option, it is possible to store data with restricted access for a limited period of time.

When research data is published, it should be assigned a PID so that the data sets can be clearly referenced (see the FAQ on PIDs). It is also advisable to assign licences for reuse (see the FAQ on licences).

Persistent identifiers (PIDs) are unique alphanumeric codes that can be used to permanently and uniquely identify digital resources. They ensure that digital objects can be found in the long term and prevent the creation of so-called ‘dead links’, for instance when a publisher’s website address changes. Assigning a PID means that a data record remains uniquely referenceable and citable.

Common PID systems for publishing of texts and research data include the DOI, Handle and PURL systems:

  • The Digital Object Identifier (DOI) is the most well-known and most widely used PID internationally. In many data repositories, including the Gutenberg Open Science repository at JGU, it is automatically assigned when research data is published.
  • The Handle system is another persistent identifier that is frequently used internationally.

ORCID enables the unique reference of authors of scientific publications. Further information can be found on the Mainz University Library website. Information events on ORCID are offered regularly.

German copyright law often does not apply to research data because copyright protection requires an element of intellectual creation. This applies to texts, images and videos, but not to scientific measurement data, for example.

Further information on copyright can be found on the Mainz University Library website.

The Mainz University Library has published a handout on this subject: https://www.ub.uni-mainz.de/sites/default/files/2019-08/Handreichung-Forschungsdatenbereitstellung.pdf

Further information can be found here: https://tu-dresden.de/gsw/phil/irget/jfbimd13/ressourcen/dateien/dateien/DataJus/DataJus_Zusammenfassung_Gutachten_12-07-18.pdf?lang=en

Copyright Act – Open Access Publication of Journal Articles

Section 38 (4): The author of a scientific contribution produced in the context of a research activity, at least half of which is publicly funded, and appears in a periodically published collection at least twice a year has the right to make the contribution publicly accessible in the accepted manuscript version after twelve months have elapsed since the first publication, even if they have granted the publisher or editoe exclusive rights, provided this does not serve a commercial purpose. The source of the first publication must be stated. Any other agreement that disadvantages the author is invalid.

This translation of the above paragraph is not legally binding.

Therefore, it is possible to publish journal articles in JGU’s institutional repository Gutenberg Open Science after one year, for example. This is relevant for fulfilling of the Open Access obligation in Horizon Europe projects, for example.

SHERPA/RoMEO

The SHERPA/RoMEO website https://beta.sherpa.ac.uk/ provides information on the copyright policies and self-archiving requirements of various scientific publishers and journals.

Licences define the conditions under which published data can be used, which can differ from copyright law. In the scientific field, free licences are often used to facilitate the reuse of data. Creative Commons licences are particularly widespread and permit the use of data to varying degrees.

Examples:

CC 0: No rights reserved

CC BY: Attribution

CC BY-SA: Attribution-Share Alike

CC BY-ND: Attribution – No Derivative Works

Further information on licences and their use in the Gutenberg Open Science repository can be found here.

Open-source software

Information on open-source software, as well as a JGU handout on licensing software under open-source licences, can be found at https://forschung.uni-mainz.de/files/2019/03/Open_Source_Lizenzen.pdf.

In the context of open data, it is often demanded that data should be FAIR. In other words:

Findable

Accessible

Interoperable

Reusable

For more information, click here:

https://www.go-fair.org/fair-principles/

https://blog.tib.eu/2017/09/12/die-fair-data-prinzipien-fuer-forschungsdaten/

If you wish to work with sensitive data relevant to data protection law, please contact the data protection officer at your institution. For JGU: https://organisation.uni-mainz.de/datenschutzbeauftragter/

When handling personal data, e.g. in the context of interviews/surveys etc., the Federal and State Data Protection Acts and the European General Data Protection Regulation must be observed.