Research data

Many definitions attempt to capture the notion of Research data. This challenge is all the more difficult because it is necessary to find a common denominator for many types of data (sociological, economic, medical, biological, constructed from sounds, images or even odours, etc.).

It is commonly accepted (OECD - 2007) that research data are factual records used as primary sources for scientific research. They are generally recognized by the scientific community as necessary to validate research results. They can take many forms (experimental data, observational data, operational data, third party data, public sector data, etc.).

At UNIL, according to Directive 4.5, research data are records used as primary sources for scientific research.

They include in particular, but not exclusively :

• primary data: original data (measurement, text, image, sound, video, questionnaires, etc.) collected or generated for the purpose of carrying out a research project ;
• existing data collected or copied for immediate or future use in Projects (in particular administrative or statistical data, digitised content from collections, data available in databases expressly made available to a research community) - original data or content is not covered by this Directive ;
• any new data resulting from the processing (analysis, aggregation, transformation, etc.) of the primary data.

The following are not considered as research data :

• files containing constituent elements of the publication of the research (texts forming the core of the publication, as well as annexes - tables, graphs, images, etc.) ;
• files generated by the project administration (scientific, financial reports) or media communication related to a project.

In order to facilitate their publication and/or sharing in possible open access, research data are organized and managed according to international standards specific to each field in order to respect the FAIR principles (Findable, Accessible, Interoperable, Re-usable) supported in particular by the Swiss National Science Foundation (SNSF) .

Research data have a long lifespan, often longer than the period between their creation and the writing of the scientific publication for which they were created. The function and value of the data changes from one phase of the cycle to the next. The concept of research data life cycle is a tool that can be used to map different phases and see how they connect to each other. The use of a life cycle makes it possible to move from a short-term perspective to a long-term perspective in data management.

Developed by UK Data Archive, the Research Data Lifecycle Reference Model  defines 6 main steps : Data creation ; Data processing ; Data analysis ; Preparing data for preservation ; Data access ; Data reuse.

Each of these steps consists of several actions to be carried out to ensure proper management of research data.

Uniris has developed a similar vision based also on 6 phases :

1. Project planning management (DMP)
2. Data collection or creation
3. Organization and analysis
4. Preservation and curation
5. Archiving and sharing (publication)
6. Reuse of data

Taking these 6 phases into account allows the following aspects to be achieved :

• Secure storage during the active phase
• Long term storage of organized and documented data
• Archiving on a FAIR and non-commercial repository, and sharing (Open research Data)

A distinction is made between active research data, the preservation of part of this data (long-term preservation) and permanent archiving and data sharing.

• active research data are data in use by the researcher ;
• long-term stored data are data that have already been analysed and are available for consultation and/or use in other research, or that have not yet been used in the first research ;
• data that are permanently archived and shared via a non-commercial and FAIR data repository are archived to allow their accessibility and reuse over time and thus meet the challenges of Open research Data.

See the diagram below.

Research data is numerous, varied and highly heterogeneous. They can be distinguished into five categories (André, 2014):

1. Observation data
2. Experimental data
3. Data from computational, models or simulations
4. Data dées or compiled;es
5. Data from reference or canonics

Depending on the context in which it is created (capture or production), how it is exploited, analysed and processed, research data can be of different kinds:

• Rough, raw, formatted, cleaned, primary, secondary, processed, etc.

Contained in various media:

• Lab notebooks, electronic documents, paper, software, computer programs, etc.

Of all types:

• Archives, audio, video, databases, source codes, geospatial, images, photographs, programming languages, material and physical, models, visualizations, 3D, digital, textual, digitized, scans, qualitative, quantitative, statistical, etc.