Research Data Management and FAIR Principles
Data Management
Research Data management is a broad term that encompasses all procedures and activities designed to ensure that research data is safe, sustainable, easily discoverable, understandable, and (re)usable not only during the research project but also in the long term.
A data management plan is a strategic document that outlines the approach and procedures for managing data during the fulfillment of a grant or for a team. It is a planned framework that determines how data will be collected, stored, processed, protected, and used. The data management plan establishes principles, procedures, and responsibilities related to data management and serves as a basic guide for all team members working with data. It is a living document that is regularly updated during the research to reflect what is happening with the data.
The data management plan generally includes several key aspects:
- Data Collection: Defines the processes and methods for data collection, including sources, collection frequency, and required data quality.
- Data Storage: Defines the ways and places where data will be stored, ensuring their availability and integrity.
- Data Processing: Describes methods and tools for data processing, including transformation, analysis, and data preparation for use.
- Data Security: Determines measures and procedures to protect data from unauthorized access, loss, or damage.
- Data Quality: Defines standards and processes to ensure data quality, including integrity checks, validation, and data cleaning.
- Metadata Management: Describes how metadata will be managed, which is information about data, such as their meaning, structure, and relationships with other data.
- Data Availability and Sharing: Establishes rules and procedures for access to data and their sharing within the organization, including permissions and sharing methods.
A data management plan is an important tool to ensure that data is managed efficiently, securely, and consistently. It helps minimize risks associated with data management and increases the value of data. Some providers have specific templates for data management plans (e.g., Horizon Europe DMP template), and there are several tools that facilitate or even simplify the completion of these templates. UCT Prague has its instance of DSW - Data Stewardship Wizard.
FAIR principles
FAIR principles are key guidelines for creating and documenting data and other outputs (including metadata): Findability, Accessibility, Interoperability, and Reusability. These principles are set to be met for both humans ("Human readable") and machine-readable ("Machine actionable") processing.
- Findability: Metadata and data should be easily findable for both people and computers (machine-readable metadata is essential for automatic data file and service discovery). For open sharing, data must be in a trustworthy repository (not just on websites!), including rich metadata with persistent identifiers and well-described. For example, the general repository Zenodo provides DOIs, digital object identifiers, to data files.
- Accessibility: Once a user finds the desired data file, they must have the ability to access it (possibly through authentication and authorization). It is necessary to know who will have access to the data in the data repository. When will data be accessible (immediately or after the embargo period)?
- Interoperability: Data is interoperable through the use of common data formats, and industry standards. Thus, data must be in open formats, according to standards for a specific field (if such standards exist). In some fields, ontological dictionaries exist that allow the use of established terms (mention them if applicable).
- Reusability: To reuse data, metadata, and data must be well-described to replicate and/or combine them in different environments. Data in the repository is provided under a license (e.g., Creative Commons CC BY). Ideally, rich metadata describes why and how data were obtained, where, when, and by whom they were collected.
Each FAIR principle does not have a metrical fulfills/does not fulfill, but rather to what extent they are met.