This page provides concise explanations of key terms relating to research data management.
Do you have any questions, suggestions for additions, or do you think certain points have been misrepresented? Please feel free to write to us – we look forward to your email.
How to Cite This Page (Chicago)
forschungsdaten.info editorial board. "Glossary". forschungsdaten.info, 01. September 2026. https://forschungsdaten.info/en/applied-rdm/glossary
A
An algorithm is a defined sequence of steps used to solve a problem. The most common example of an algorithm is a recipe: the steps can be described unambiguously, can be carried out in a hands-on fashion, and always produce the same result when followed exactly. In principle, an algorithm can be formulated in any language (natural language or programming language). Generally speaking, virtually all automation processes are based on algorithms. In computer science, algorithms form the basis of almost all programmes and applications. They are used, for example, when structured data (such as authority files or metadata) or unstructured data (e.g. in data mining) is to be recorded in a standardised manner. In the context of machine learning or artificial intelligence (AI), algorithms are used to recognise patterns in large amounts of data and to abstract complex relationships.
Analogue research materials are physical objects, such as samples, physical photographs, handwritten notes, books, audio tapes, paintings, fossils, drill cores, or architectural models. To make analogue materials usable in a repository, they must either first be digitised (digital copy) or, at least, the associated metadata must be made available digitally. Analogue materials are the opposite of born-digital data, as the latter were created in a digital form and have thus always been digital.
The anonymisation of personal data in academic research is part of good research practice. According to paragraph 36 of the DSG NRW (Data Protection Law of North Rhine-Westphalia), anonymisation refers to any measures that alter personal data in such a way that “the individual details regarding personal or factual circumstances can no longer be attributed to a specific or identifiable natural person, or can only be attributed to such a person with a disproportionate expenditure of time, cost and labour” (translation of the original quotation). A distinction must be made between anonymisation and pseudonymisation.
An archive is generally understood as a collection of documents that are retained indefinitely. Archives are provided as service offerings, but they can also be established and operated under one’s own responsibility. The content’s organisation is based on the guidelines that underlie the respective archive. In research data management, the term archive refers to a collection of data. For research data, an archiving period of at least ten years has become the norm. A common form of archiving research data is done in repositories. A specialised form is the so-called long-term archiving.
For security or confidentiality reasons, some data, systems or services require access restrictions. In this case, the so-called authentication regulates the access. Via authentication, the legitimacy of individual access attempts can be verified with a number of different characteristics, such as IP address, login and password, security feature (key file, biometric feature, hardware token) or a combination (two-factor authentication). A functioning user management system or identity management (IDM) system for storing and managing the aforementioned characteristics is a prerequisite. One variant of authentication is the so-called single sign-on procedure, which is used by, e.g., Shibboleth. It allows a person to use multiple services with a single login. Authentication must be distinguished from authorisation, in which specific rights within the system are assigned to a specific person.
B
The term “best practice” refers to an optimal solution, an ideal method, or at least a well-established approach to a (work) process. Generally, this “best practice” has evolved over time with experience. When compiled as a “best practice example,” this approach serves as a model explicitly recommended for replication.
Digital data consists of a defined sequence of bits, which have the value 0 or 1 and are stored on data carriers. This sequence is called a bitstream. Bitstream preservation is the ability to preserve the bitstream beyond technological changes. This is a fundamental prerequisite for digital long-term archiving.
The blockchain is a digital technology used to make the documentation of information traceable and tamper-proof. In research data management, especially the so-called permissioned blockchains are used; these allow only known and authorised parties to access or add to specific stored information.
In addition to transparently documenting the origin and processing of research data (data provenance), permissioned blockchains permanently record research-related workflows in a traceable manner. Through cryptographic methods, immutable ledgers and automated rules (smart contracts), the data’s integrity can be guaranteed; access rights and collaborations among research partners can be supported. The blockchain thereby contributes to traceability, good research practice and the reproducibility of research-related workflows.
The term ‘born digital’ refers to data and materials that were created in a digital form. This term is not applicable to analogue materials which have been digitalised some time after their creation.
C
The CARE Principles for Indigenous Data Governance were developed by the Research Data Alliance International Indigenous Data Sovereignty Interest Group as a complement to the FAIR principles and published by the Global Indigenous Data Alliance (GIDA).
The acronym stands for Collective Benefit, Authority to Control, Responsibility and Ethics. Based on the CARE Principles, researchers are made aware of the need to safeguard the rights and interests of indigenous communities in efforts towards Open Data and Open Science; regardless of whether their research specifically concerns the communities themselves or affects them in other ways. The aim of the CARE Principles is, among other things, to make sure that the right of self-determination of indigenous individuals and groups is not disregarded due to unequal power relations or historical inequalities, e.g. through the use of data without taking the data providers into consideration. Moreover, the CARE Principles call for data sovereignty of indigenous communities. This also applies to data and other research findings relating to natural resources (e.g. georesources or plants) on the communities’ lands.
Closed Access refers to the access to published content that is only available via a barrier (e.g. paywall, subscription). Unlike Open Access, Closed Access usually denotes a conventional publication format that involves subscription fees charged by publishers or providers.
Data can be stored and shared in clouds, independently from local computers and networks, thereby enabling collaborative work on data. As clouds can be accessed from anywhere, they are a key component of research work involving data – for both individual researchers and research groups. Furthermore, software applications, computing power and other IT services can be obtained via cloud providers. Some institutions (e.g. companies) operate self-managed, private clouds; some IT companies offer standardised public cloud services. The use of such solutions raises questions regarding data security and data protection, particularly when the services are located outside the EU. For researchers, it is therefore recommended to use institutional or regional cloud services such as Sync&Share, Sciebo, Academic Cloud or Switch Drive.
Cold data is research data that is accessed only very rarely. Due to that, it does not require short retrieval times, meaning that higher access latency can be accepted. It usually consists of finalised datasets that are no longer modified and thus remain largely unchanged for years or even decades. Cold data is primarily used for archiving purposes, for example for documentation purposes, to ensure reproducibility, and to meet legal and institutional compliance requirements. Together with descriptive metadata, it is stored in repositories for publication and/or archiving. For long-term storage, object storage, RAID arrays with HDDs (Redundant Array of Independent Disks) and tape drives (long-term archiving) are primarily used. The latter offer particularly cost-effective storage over long periods.
To ensure the maximum reusability of research data, it is recommended to actively grant usage rights, e.g. through licences. The use of liberal licensing models, in particular the globally recognised Creative Commons licences (CC), is one way of establishing transparent and legally sound conditions for the re-use of published research data regardless of differences in national copyright law
D
A Data Curation Profile describes the ‘history’ of a dataset or data collection, i.e. the origin and lifecycle of a dataset within a research project. The profile developed by Purdue University Libraries, along with the associated toolkit, is both a tool and a full-fledged collection of datasets. The tool is an interview instrument based on a template that facilitates a very thorough “data discussion,” which creates the “profile” as it is completed. The data collection can be searched for completed Data Curation Profiles, for example to support research data management units in the data curation within a specific discipline or research method.
Read the article
Published data must be described in sufficient detail to ensure meaningful re-use. Data Description Templates are pre-structured documents that list the necessary information and standardise the structure of data descriptions. They thus simplify the creation, curation and reading of data descriptions.
Data literacy describes the ability to handle data systematically throughout the entire data lifecycle. This encompasses the collection and selection of data, as well as its processing, analysis, presentation and sustainable use. Beyond that, data literacy includes a critical reflection on the handling of data (e.g. societal issues).
Data stewards are experts in research data management. They are primarily employed at research institutions to support researchers in the sustainable handling of data. The tasks of data stewards primarily include consultations, training sessions, needs assessments, and requirements management. Decentralised data stewards work at faculty, institute or project level and assist researchers with discipline-specific concerns. Data stewards are often trained through RDM-related professionalisation programmes (e.g. certificate courses).
File formats (sometimes also referred to as file type or data format) are generated when files are saved; they contain information about the structure of the data contained in a file, as well as its purpose and affiliation. Based on the information contained in a file format, application programmes are able to interpret data and make its contents available. File formats can usually be identified by a corresponding extension appended to the actual file name. Such extensions consist of a dot followed by two to four letters.
In the case of so-called proprietary formats, files can only be opened, edited and saved with the corresponding application, utility or system programmes (for example, .doc/.docx, .xls/.xlsx). Open formats (for example, .html, .csv, .jpg, .mp3, .gif), on the other hand, allow files to be opened and edited with software from different manufacturers.
File formats can be actively changed during the saving process by means of conversion, which may result in data loss. In academia, file formats should be chosen with particular attention to compatibility, suitability for long-term archiving and lossless conversion to alternative formats. Therefore, open file formats should be preferred, wherever possible, when publishing and archiving data.
Read the article
The German database protection law (Sections 87a–87e of the Act on Copyright and Related Rights or Copyright Act) protects a database against unauthorised use and reproduction for a period of 15 years, provided that the database’s creation required a ‘substantial investment’ of money, time, labour, etc.. It is not the reproduction of the entire database that is prohibited, but only that of a “qualitatively or quantitatively substantial part”, although it may not always be clear how this is defined.
Database protection law exists in the EU, but not in Switzerland. It serves to protect investment. Although database protection law functions similarly to copyright law and is also found (at least in Germany) in the Copyright Act, it does not constitute copyright protection in the strict sense. Consequently, it is not necessary for the database to attain the so-called “intellectual level of creativity”; only a ‘substantial investment’ is required.
In addition, a database – in both Germany and Switzerland – may also be protected by copyright as a compilation if it has individual character and constitutes an intellectual creation (in Germany, this is referred to as a “database work” as a sub-category of a compilation). Furthermore, the individual data records may also be protected by copyright.
Data journals are a publication venue for datasets. They provide comprehensive descriptions of published datasets without interpreting them, thereby supporting the re-use of research data and their recognition as achievements in research. At the same time, they aim to improve the transparency of research methods and results and thereby support good RDM practices.
Read the article
Data journalism refers to a form of journalism which systematically researches, collects, processes and analyses data to produce journalistic content. The results are often presented via tables, diagrams, maps or interactive visualisations. Data journalism thus combines journalistic research with data analysis and can draw on both publicly available data and research data.
Data journalism should be distinguished from data journals: whilst data journalism refers to a journalistic form of working and presenting information, data journals are academic publication venues in which datasets are described and documented.
Data competence centres are dedicated to teaching data literacy to researchers across disciplines. With a predominantly regional focus, they develop and provide educational resources, establish support services, and organise events and platforms for exchange and networking. All of this aims at shaping the cultural shift regarding the use of research data.
Data competence centres should not be confused with data centres and research data centres.
Overview page
Data curation encompasses all actions necessary to select research data, to maintain it (sustainably), to make it reusable and to preserve it in the long term. The curation process begins with the selection of data for storage based on their relevance for research, their quality or legal requirements to store them. In the broadest sense, curation also refers to a series of measures to manage, maintain and validate this data in the sense of active and ongoing administration throughout the lifecycle.
Data curation is a central requirement for locating and retrieving data, as well as for maintaining its quality, its research-related value and its potential for future reuse.
The data lifecycle model encompasses all phases research data may go through, from the planning stages of data use to data reuse. The data lifecycle’s structure varies from model to model, but generally comprises the following phases:
- Planning the work with and the handling of data (see also Data Management Plan)
- Collecting data
- Processing and analysing data
- Sharing and publishing data
- Archiving data
- Re-using data
Read the article
Data management plans (DMPs) describe the handling of research data produced or (re)used within a research project. DMPs not only cover the data processing stages of a research project, but also extend beyond it. They align with the research data lifecycle, but also involve legal and ethical frameworks as well as personnel responsibilities. DMPs thus support the systematic planning and implementation of data management measures, as well as the calculation of the material and human resources required therein.
DMPs support the planning stages of RDM and help to describe the handling of data in a traceable and structured manner. Furthermore, they help to structure and standardise the collaborative work with data within a research group. Accordingly, DMPs contribute to the quality assurance and long-term usability of research data and support their users in implementing the FAIR principles. DMPs are living documents and should therefore be updated regularly.
Numerous research funding bodies now require a DMP or at least a section of text relating to RDM as a condition for funding. However, an initial version of the DMP is usually only required at or shortly after the start of the project.
Read the article
Data mapping is the process of transferring data (or data elements) from one data model or metadata schema to another. This is the first step towards integrating external information into an information system. It encompasses data transformation when data is exchanged electronically. Here, the markup language XML and the data format JSON are very frequently used.
A law in Germany that stipulates how data held by public bodies (including research data from publicly funded research) must be published. The rules do not require sharing, but merely regulate the ‘how’ once someone has decided to share data or is obliged to do so, for example, by a funding agreement or research data guidelines.
The term is used ambiguously and can refer to:
- An agreement on the transfer and re-use of research data that is not covered by an open licence (Data Transfer and Use Agreement, DTUA). Particularly in the case of sensitive research data, such as in medicine, the use of open licences is often not possible. Through a DTUA, the data can nevertheless be made accessible on a case-by-case basis.
- An agreement (contract) between the researchers involved in a research project, which establishes as early as possible (ideally before data collection begins) who will have what permission in handling the data and at what point in time. The agreement can and should also regulate how the data will be reused and/or published after the project ends. This helps to prevent conflicts that might arise from subsequent disagreements over the best possible handling of the collected research data.
- Occasionally, the term is also used to refer to the data protection consent form on websites. This use of the term has nothing to do with research data and is included here solely for the sake of completeness.
Data protection refers to technical and organisational measures against the misuse of personal data collected, for example, in medicine or social science research. Misuse occurs when such data is collected, processed or used without authorisation. Data protection is regulated, among other things, by the EU’s General Data Protection Regulation (GDPR) and in the relevant provisions of the respective countries.
Appropriate data protection measures include encryption techniques or the storage of data in specially secured locations. Furthermore, subsequent anonymisation or pseudonymisation can remove personal references to such an extent that publishing this data may, under certain circumstances, become legally permissible.
The deliberate, redundant storage of data is referred to as data backup. The aim is to ensure the recoverability of the backed-up data’s status (e.g. of raw data) in the event of data loss due to, for example, a device defect (water damage, hacking), the loss of equipment (USB stick, laptop) or errors made during data cleansing. Backups are recommended during all phases of the research data lifecycle, particularly during the collection and analysis phases. Ideally, backups should be created automatically and regularly. The 3:2:1 rule is recommended: three copies (working file and two copies), on two different types of media (e.g. hard drive and cloud), with one copy stored at an off-site location (e.g. in a cloud with distributed servers).
Read the article
Data trust centres or data donation centres play a particularly important role in the re-use of personal or sensitive data. As a rule, this data must not be freely accessible either for legal reasons or, in some cases, the data should not be available for ethical reasons. This serves to protect the personal rights of the data providers, particularly where (complete) data anonymisation or pseudonymisation is not possible, as this would result in the loss of important information. Data trust centres or data donation centres are tasked with controlling access to (personal) datasets. Upon request, researchers can, for example, obtain access to anonymised or pseudonymised datasets, or to sections of datasets relevant to their research, which would otherwise not be possible. This is also conceivable, for example, when working with corporate datasets or data collections whose contents should not be openly accessible for ethical reasons (e.g. hate speech, property of indigenous communities, etc.). At the same time, research findings can be fed back to data providers via data trust centres (e.g. regarding personal risks that have been identified by the research).
In specialist disciplines such as medicine, data trust centres are well established. Along these lines, the legal framework has been adapted accordingly to enable data donations from patients (see the Health Data Use Act (GDNG)). In other disciplines, data trust centres are currently in progress.
A data centre is a facility used to consolidate IT infrastructures. The term refers both to premises housing servers and network technology, and to locations where data is managed, stored and analysed. Data centres exist in both research and the private sector.
In research institutions, data centres are usually full-fledged departments, or at least departments closely linked to the institution, that are responsible for the management and storage of data collected either by the institution itself or by its cooperation partners.
Data centres should not be confused with data competence centres and research data centres.
Digital Object Identifiers (DOIs) are one of the most common systems used to persistently identify digital objects. Specifically, a DOI is an online reference assigned to a digital resource (e.g. an article in a journal or a research dataset) to provide it with a unique and permanent reference on the internet. A DOI remains the same throughout the entire lifetime of the object it identifies – regardless of changes to websites or server shutdowns. The DOI system is administered by the international DOI Foundation.
E
Electronic Laboratory Notebooks (ELNs) play a central role in documenting the planning, execution and evaluation of experiments in research – and are therefore also key tools in research data management. They often replace paper-based laboratory notebooks, but their functionality also extends beyond that of their analogue predecessors. In addition to generic ELNs, there are discipline-specific ELNs that cater to the needs of individual disciplines.
Read the article
In research data management, a (time-limited) embargo defines a period during which only descriptive metadata of research data is accessible. The data associated with the metadata is not accessible during the embargo. An embargo may be used when research data is supposed to be published with a time delay (e.g. in the context of peer-review processes).
In academic publishing, an ‘enhanced publication’ refers to an electronic publication which, in addition to the publication text, also contains associated materials such as research data, videos, tables or presentations.
F
The DFG’s Discipline-Specific Information Services Programme (in German: Fachinformationsdienste (FID)) aims to strengthen and improve research-related information infrastructures. It is the successor programme to the Special Collections (Sondersammelgebiete) funding programme. Many FIDs offer discipline-specific services on research data management.
Read the article
A FAIR Digital Object (FDO) is a record that bundles all essential details about a (research) dataset, thereby creating a machine-processable, meaningful and technology-independent object. Based on the FAIR principles’ abstract guidelines for handling (research) data, FDOs define concrete technical concepts to implement the FAIR principles on an operational basis.
FDOs represent a central concept for the automated use of research data, as they link the dataset (bit sequences) via a unique Persistent Identifier with machine-readable information (metadata and context) as well as operations. This self-describing structure enables machines or agents to process the data further without human intervention.
For the technical implementation of FDOs, persistent identifiers (e.g. handle-based PIDs) and a core dataset (Kernel Information Record) are stored directly within the data’s digital identity. This core dataset comprises machine-readable information such as technical metadata, e.g. object types, API endpoints, as well as details on rights and licences. Machines can retrieve this information directly and process it automatically without human intervention. Through defined operations in the API endpoints, FDOs can also describe datasets with complex structures and queries (e.g. SQL). In research projects, this approach promotes interoperability between different systems and facilitates the seamless exchange and sustainable reuse of research data. This approach is implemented, for example, by Coscine to describe projects and resources in an interoperable manner.
The main objective of the FAIR principles is the optimal preparation of research data, which, in accordance with the principles, should be findable, accessible, interoperable and reusable. Central to this is the description of research data with metadata as well as adequate documentation. Even research data that is not publicly accessible can comply with the FAIR principles, provided, for example, that the data’s metadata is published. The FORCE 11 community first coined the term ‘FAIR Data’ in 2016 to promote sustainable research data management. The FAIR principles now form a central basis for RDM requirements set by funding bodies such as the European Union or the Swiss National Science Foundation.
Read the article
Research data are digital or analogue data generated, developed or analysed in research (e.g. through measurements, surveys, source work) or data based on such. They are a basis for research and document research results. What exactly the term ‘research data’ encompasses varies from discipline to discipline.
The term ‘research data management’ refers to structured measures in the context of working with research data. Among other things, these measures aim to ensure that data can be used or reused in the long term independently of the individuals involved in the data’s collection. This increases the efficiency of research (e.g. in the context of a research group’s work, but also in terms of the global research progress). A further objective is the implementation of legal requirements and ethical best practices in the handling of sensitive data, such as personal data. Research data management encompasses not only the publication of data (Open Data), but also measures related to the preceding stages of the entire data lifecycle, as well as data archiving and reuse.
A generally applicable institutional regulation (at universities often in the form of a guideline), aimed at researchers and governing the handling of research data during and after a research project in a more or less binding manner. A policy’s exact content, legally binding nature and length can vary greatly across institutions.
Read the article
Research data centres are places where data is archived and made accessible for broad use in research. They play a central role in e g. social, behavioural, educational and economic research. In addition to data from studies performed in the aforementioned disciplines, they focus primarily on geodata and spatial data, financial data, register and social security data, as well as statistical data.
Research data centres should not be confused with data centres or data competence centres.
G
Good Research Practice refers to steps and measures in research designed to ensure that research is conducted in an appropriate fashion. In Germany, the German Research Foundation’s guidelines for safeguarding good research practice [German only] are particularly essential in this regard, as research institutions can only receive funding from the German Research Foundation if they implement these guidelines in a legally binding manner. Many of the guidelines have an impact on the handling of research data in the context of research data management.
Read the article [German only]
H
Harvesting refers to the systematic and strategic browsing of websites and content by bots. In the context of research data management, harvesting primarily refers to the automatic ‘collection’ of data or metadata directly from archives and repositories or via so-called data providers (such as BASE and OAIster).
In this process, harvesting protocols are used to automatically retrieve data. In many cases, documented application programming interfaces (APIs) are made available. One of the most commonly used harvesting protocols for metadata is the XML-based Open Archives Initiative Protocol for Metadata Harvesting (OAI-PMH). As there is a multitude of very different metadata standards, the Dublin Core model uses harvesting via OAI-PMH as the lowest common denominator for metadata representation. See also: text and data mining [German only].
High-performance computing (HPC) refers to a field of computing dedicated to handling complex or large-scale calculations and data analyses. This typically places high demands on computing power and storage capacity; the parallel processing of computational tasks also plays a central role. Appropriately equipped computing clusters rely on the fastest possible connectivity and extremely short response times between the respective computer units.
Read the article [German only]
Hot Data refers to research data that is accessed frequently during an active research project. Due to that, it must be immediately available for processing in order to enable high-performance analyses, even of large data sets. During this phase, data is processed intensively and subject to continuous changes. It is therefore in a dynamic state and not yet intended for publication. Data access is usually managed within a project team, granular access and audit rights are crucial. Fast flash-based technologies, such as solid-state drives (SSDs) or RAM, and in some cases specialised high-performance object storage systems, are used to store hot data. Whilst these storage solutions ensure high access speeds, they come with higher costs and consume more energy than other solutions. They should therefore only be used when such high speeds are absolutely essential. Due to the intensive use of these storage media when working with hot data, a higher replacement rate for these media is necessary.
In environmental studies, water, air and soil samples are taken, for example to determine biodiversity using environmental DNA (eDNA). In the process, human genetic material is collected unintentionally. This human genetic material is referred to as human genetic by-catch, as the focus is usually on detecting animals, plants and fungi. In addition to ethical concerns, there are concerns regarding data protection, as human genetic by-catch can be used to prove that a specific person was present at a specific location.
A critical examination of the topic was published by Whitmore et al. in Nature in 2023.
Whitmore, Liam, Mark McCauley, Jessica A. Farrell, et al. 2023. “Inadvertent Human Genomic Bycatch and Intentional Capture Raise Beneficial Applications and Ethical Concerns with Environmental DNA”. Nature Ecology & Evolution 7 (6): 873–88. https://doi.org/10.1038/s41559-023-02056-2.
I
Informed Consent refers to a process in which researchers inform their research participants (e.g. interviewees) about the project in which they are involved. This includes the data collected in the project as well as the question of what happens to this data. On this basis, participants should be able to decide voluntarily and in a well-informed fashion whether to accept or decline their participation. Obtaining informed consent is not only a legal requirement under Article 6 of the GDPR but is also required by research ethics when personal data is collected in a research project. At the same time, informed consent (in most cases) also refers to a written document which is to be signed by the participants. In some cases, however, obtaining written informed consent can be difficult, e.g. in particularly sensitive research contexts where participants do not wish to or are unable to provide a signature. In such cases, verbal consent (without a signature, but documented as thoroughly as possible) may be considered as an alternative. Procedural consent (repeatedly obtaining consent throughout the course of the project) may also be advisable, e.g. if the aim is to ensure a particularly close involvement of the participants in the research process. Templates and recommendations regarding the specific content of a document or a script for informed consent are available, for example, at the RDC Qualiservice. The requirements related to informed consent are also set out in Article 7 of the GDPR.
Ingest refers to the process of adding research data to an archive or repository. The technical ingest – i.e. the simple upload of data into the storage system – may be accompanied by quality controls (e.g. the selection or adaptation of the data and file formats, the verification of metadata or confidentiality) and various preparation steps (metadata enrichment, etc.). The workflows vary from archive to archive, from repository to repository and depending on what content is to be ingested.
Interoperability generally refers to the interaction of different systems which, under ideal conditions, runs seamlessly (or as seamlessly as possible), for example to enable that data can be exchanged efficiently, interpreted by machines and combined automatically with other datasets. In the field of research data management, interoperable metadata play a central role alongside compatible file formats. Interoperability is one of the core concepts of the FAIR principles.
J
JSON is a compact, easily readable and software-independent data format used to exchange data between applications. It is used in particular in web applications for the transfer of structured data, in order to integrate this data into other systems or applications. For the same information, JSON requires significantly less storage space than XML, but is not as versatile.
K
A controlled vocabulary is a standardised form of language in which a word or phrase has exactly one defined meaning. Controlled vocabularies are used, for example, when assigning keywords (known as ‘subjects’ or ‘topics’) in metadata to describe digital objects. They are also often stored in authority files (e.g. The Integrated Authority File (Gemeinsame Normdatei, GND)), thesauri (e.g. Getty Arts & Architecture Thesaurus) or indexes. A key function of controlled vocabularies is to link synonyms (e.g. cell phone/mobile phone) which lead to the same search result, as well as distinguishing between different meanings of a term (e.g. a ‘bank’ along the sides of a river versus a financial institution). In the context of the FAIR principles, controlled vocabularies play a central role because they enhance the findability, interoperability and reusability of research data. Examples of controlled vocabularies include the Data Documentation Initiative (DDI) Controlled Vocabularies or the Getty Arts & Architecture Thesaurus (AAT). Controlled vocabularies can be found in databases such as BARTOC or FAIRsharing. Ontologies go a step further than controlled vocabularies, as they can be used to describe the relationships between terms in a nuanced way.
Artificial intelligence (AI) refers to a subfield of computer science in which, among other things, human decision-making structures are replicated with computational methods. This enables machine applications to perform cognitive tasks such as learning, judgement and problem-solving via algorithms, all of which can also be utilised in research data management. The term is difficult to define due to the lack of a clear definition of ‘intelligence’.
L
In Germany, a state initiative is a network or project supported or funded by a federal state to assist universities, other research institutions, and non-academia state institutions in handling research data. The state initiatives’ objectives include the networking of institutions within their state, the provision of information, and the offering of consultations, trainings, and support on research data management. State initiatives often coordinate RDM activities within a federal state and support the implementation of national or state-wide RDM strategies.
Read the article
Long-term archiving aims to provide access to archived data over long periods of time. In the field of research data management, a retention period of at least ten years has become the standard for research data. The idea is to prevent data loss and to ensure the data’s long-term authenticity. To this end, it is essential to use a suitable archiving system.
Read the article
Linked Open Data is structured, machine-readable data which has been published on the internet under an open licence. It uses URIs to identify resources and to enable links to other datasets.
A licence is a contractually agreed right of use. Through it, rights holders (licensors) permit their contractual partners (licensees) to use a work in various ways (e.g. copying or storing it, making it digitally accessible). In many cases, licensors charge a fee. In addition to such commercial licences, there are free licences such as the Creative Commons licences. These permit the use of a work free of charge.
Read the article
M
Machine-readable data can be found and used automatically by computer systems with little or no human intervention. A prerequisite for a data’s machine-readability is the use of a uniform structure. Machines or computers that are intended to read and use this data are programmed on the basis of this structure.
Metadata comprises structured information about data (e.g. research data) or other resources and their characteristics. It is stored either independently of or together with the data it describes. A distinction is usually made between subject-specific (e.g. measuring instruments used) and technical or administrative (e.g. DOI) metadata. Whilst the latter have a distinct status as metadata, subject-specific metadata can sometimes also be regarded and used as research data. To increase the effectiveness of metadata and that of the dataset it describes, the use of metadata standards is essential. The use of standardised metadata improves the findability and reusability of the data or resources associated with the metadata. Standards thereby enable metadata and data from different sources to be linked with each other and processed together.
Read the article
Metadata schemas organise the structure of metadata. They specify which elements are used to describe objects such as research data and what information should be provided in what form. Standardised metadata schemas harmonise the data collection process and enhance the metadata’s interoperability and quality. Furthermore, structured metadata ensures long-term reusability. It also enables machine readability and the exchange of information between different applications.
Metadata standards have been defined to ensure metadata interoperability, i.e. the linking and joint processing of metadata. They are used to describe similar data in a uniform fashion both with regards to content and structure. Many metadata standards can be converted into other ones via so-called mappings.
Read the article
N
The National Research Data Infrastructure (NFDI) aims to systematically catalogue research data sets, interlink them, and make them available for sustainable and high-quality use. Along these lines, the NFDI’s objective is the strategic expansion of research data management in Germany. This is achieved, among other things, through subject-specific consortia that develop service portfolios, such as consultancy offerings, for their disciplines. Interdisciplinary topics are addressed in the NFDI’s sections. The NFDI is organised as the non-profit association Nationale Forschungsdateninfrastruktur (NFDI) e.V., based in Karlsruhe.
Read the article
A Non-Disclosure Agreement (NDA) – also known as a confidentiality or secrecy agreement – is a legally binding agreement between two or more parties that determines the confidential handling of non-public information and restricts its disclosure to third parties.
When managing research data in projects that involve collaborative partners (e.g. research institutions and commercial enterprises), NDAs are used unilaterally or mutually to determine the use of data generated during a project or provided by the partners for a project. The aim is to define the circumstances under which data is exchanged, as well as to clarify how data is used after the project has concluded.
Authority data are semantically structured, controlled and maintained datasets used to unambiguously name and link content (e.g. persons, keywords, corporate bodies, works). They resemble persistent identifiers in this respect, but are more extensive in terms of content. Beyond that, they are created by trained curators, and provide, for example, name variants or links to personal relationships, places (of activity), biographical data or superordinate and subordinate terms. The Virtual International Authority File (VIAF) aims to link various authority files that are used on an international scale.
O
The XML-based Protocol for Metadata Harvesting (PMH) of the Open Archives Initiative (OAI) supports the automated harvesting of metadata via so-called data and service providers by making the data stored in repositories more accessible. OAI-PMH is widely used and easy to implement. As a standardised communication protocol, it is open and implementable, and thus complies with the FAIR principles.
Licences are standardised contracts that allow authors to legally regulate terms of reuse. In the research data context, they are frequently used for data and software. A common model are the Creative Commons licences; apart from that, e.g. the Open Data Commons licences or the German data licence (Datenlizenz Deutschland, German-language only) are tailored to data specifically. According to the Open Data Handbook, the only conditions underlying the concept of open data are those requiring the source to be cited and the data to be shared under the same terms. When applied to the CC model, this means that the CC0, CC BY and CC BY-SA licences qualify as open, the others do not (see also Conformant and Non-Conformant Licenses). Apart from that, it should be noted that quantitative data is often not eligible for protection at all, meaning that its reuse needs no permission. In this case, the use of a restrictive licence would create the wrong impression that the content is protected. In the software sector, licences such as the MIT License or GNU GPLv3 are very permissive.
An ontology is a formal, machine-readable description of technical terms and their relationships within a specific domain. It defines standardised terms, concepts and relations, thereby forming the basis for semantic interoperability, reusability and the automated processing of research data. Unlike thesauri, which provide controlled vocabularies for indexing, an ontology defines classes, relations, instances and axioms within a formal logic, thereby enabling, for example, consistency checks and automated processing.
Open Access refers to the free-of-charge and ideally barrier-free access to digital research content. As a rule, users are granted extensive rights of use and straightforward access. Depending on the definition, Open Access encompasses various forms of publication such as texts, research data (Open Data) or educational materials (Open Educational Resources); other definitions refer exclusively to text publications. Open Access allows research information to be disseminated, reused and further processed to the greatest possible extent. In the case of text publications, a distinction is usually made between Gold Open Access (publications in Open Access media such as Open Access journals that charge publication fees), Green Open Access (self-archiving via, for example, repositories) and Diamond Open Access (open-access publications in, for example, open-access journals that charge no publication fees). Detailed information on Open Access is available on the portal open-access.network.
Open Data refers to freely accessible data on the internet that may be (re-)used by third parties. Usage-related restrictions exist solely to preserve the data’s origin and open nature. For example, in the case of copyright-protected data, the use of Creative Commons licences may make naming the author a condition of data reuse. The aim is to offer a high degree of transparency and to promote collaborations through free re-use. In contrast to Open Research Data (ORD), the term Open Data is commonly used synonymously with Open Government Data (open administrative data from, for example, public authorities).
The term Open Government Data (OGD) describes public sector datasets that, in the spirit of Open Data, are freely accessible and reusable on the internet under open licences. OGD can be used, for example, in journalism or in economic contexts for the development of new products, services and business models, as well as in academia and by the civil society to promote research and public welfare. In accordance with open government strategies at, for example, the national level, open government data is proactively made available by public authorities via their own repositories.
Open Research Data (ORD) is research data that, in the spirit of Open Data, can be reused and redistributed without legal or technical restrictions. In accordance with good research practice, the creators should be credited when the data is reused, even if the data is not formally protected by copyright. Traceability, transparency and reproducibility of research results are often cited as reasons for the call for open research data. Furthermore, open research data is intended to promote collaborations within academia and beyond (e.g. with the private sector), as well as to strengthen the link between research and civil society. Although FAIR Data and open research data are not essentially synonymous, there is an overlap between the two concepts; thus, alongside FAIR Data, the meaning of ORD is often expressed as ‘as open as possible, as closed as necessary’ to account for legitimate access restrictions, for example in the case of personal data or internal company information. ORD is made available via repositories and can be found via meta-search engines such as BASE or OpenAIRE Explore.
The term Open Science encompasses principles and practices of research that aim (in the context of digitalisation) to make as many steps as possible of the research process freely accessible and reusable for researchers and society. These include, for example, access to research publications (Open Access), the development and provision of research software (Open Source), the free availability of data from administration, business and research (Open Data), and free materials for (higher education) teaching (Open Educational Resources, OER). Beyond that, traditional forms of doing research are also undergoing changes as a result of the Open Science concept (e.g. Open Peer Review, Citizen Science).
In accordance with the principles of Open Science, the production of and the access to research-related knowledge should be inclusive, equitable and sustainable. The goals of the Open Science movement are to improve transparency and reproducibility, to increase research efficiency, to democratise knowledge and to ultimately generate added value for society.
Open Source is a term used for software whose source code, under certain license terms, can be viewed, used and advanced free of charge. On platforms such as GitHub, open source code can be developed collaboratively, thereby pooling expertise, saving or sharing development costs.
The Open Researcher and Contributor ID (ORCID) is an institution-independent, internationally recognised persistent identifier used for the unique identification of individuals. Any individual can create an ORCID for themselves and maintain information in their profile, such as a CV or a list of publications. Each ORCID consists of 16 digits in four blocks of four (e.g. 0000-0002-2792-2625). ORCIDs are established at many research institutions, publishers and other research-related organisations. Along these lines, they are, for example, an integral part of the peer review process for text publications.
P
Persistent identifiers are permanent digital identifiers that consist of digits and/or alphanumeric characters and are assigned to individuals, research objects, or other entities (e.g. text publications, datasets, samples, organisations) and refer directly to them.
Common identifiers are the DOI for publications, the ORCID for individuals and the ROR for organisations. Unlike other identifiers, such as URL addresses, persistent identifiers refers to objects themselves and not to its location on the internet. If the location of a resource associated with a PID changes, the identifier remains the same. This ensures that an object remains permanently findable, accessible and citable. The central point of contact for persistent identifiers is the PID-Network.
Read the article
The EU’s General Data Protection Regulation (GDPR) defines personal data as “any information relating to an identified or identifiable natural person (‘data subject’); an identifiable natural person is one who can be identified, directly or indirectly, in particular by reference to an identifier such as a name, an identification number, location data, an online identifier or to one or more factors specific to the physical, physiological, genetic, mental, economic, cultural or social identity of that natural person” (Article 4(1)).
The Swiss Federal Act on Data Protection (FADP) provides a definition similar in substance and defines “personal data: all information relating to an identified or identifiable natural person”. (Article 5(a))
Platform-as-a-service (PaaS) refers to the cloud-based provision of runtime environments in the context of IT development, usually for a fee. The advantages and disadvantages of PaaS for organisations and their members are similar to those of Software-as-a-Service (SaaS): outsourcing the provision and maintenance of the required IT infrastructure to external service providers, as well as the potential for dependency on these providers, the need for a permanent and stable internet connection, and possible hurdles in handling sensitive data.
Pre-registration refers to the description of a research methodology which a researcher or a research group plans to use in an upcoming scientific study. Pre-registration makes public which methods are to be used to achieve the study’s objective to ensure transparency, replicability and quality assurance in research.
Primary data (also known as raw data or original data) refers to data that has been obtained directly in the data collection phase or gathered from investigations or observations (e.g. unprocessed, unchecked or unannotated measurement data or audio and video recordings). Data derived from or processed from primary data is referred to as secondary data.
Proprietary means that something is owned by a company, person or organisation. Proprietary items are usually not freely accessible; they are subject to usage restrictions, and their use depends on the provider. This often applies to software, file formats or interfaces. Proprietary file formats are particularly problematic for re-use and archiving and should therefore, where possible, be replaced by open-source alternatives.
A checksum is a (numerical) value used to verify the integrity of a digital object. The algorithm used to calculate the checksum is chosen such that any alteration to the digital object also results in a change to the checksum. If the digital object is copied, the checksums of the original file and the copied file can be calculated and compared. If the checksums differ, an error was made during the copy. A simple form of checksum is the check digit in the IBAN or the ISBN. The checksums for files - which are calculated, for example, by data repositories to verify the integrity of the data - are usually based on so-called cryptographic hash functions applied to the bitstream. These functions respond to even the smallest changes in the source data with almost entirely different checksums. An example of such a hash function is the MD5 function, which is used by Zenodo or RADAR.
The pseudonymisation of personal data means that direct or indirect indicators (e.g. name, age, place of residence) are replaced by pseudonyms (e.g. letter and/or number codes) to make it more difficult or impossible to identify the individuals to whom the data relates. Detection files, in which the original values are linked to their respective replacement values, allow pseudonymisation to be reversed. To ensure the protection of research participants, it is essential to store pseudonymised data and detection files separately when performing research data management. Unlike anonymised data, pseudonymised data remains subject to data protection regulations, as personal references can be restored via pseudonymisation keys.
R
ReadMe files contain concise and structured information on digital media such as research data and software to assist with the understanding of these media and their use. To this end, ReadMe files, for example, collate key metadata, provide information on naming conventions used, list folder structures, abbreviations and standardised data, as well as record changes to and versions of research data. They are often available as simple text files or as XML files (.txt; .md; .xml) and support their users in storing research data in a structured way. They can be archived and published with the data they describe.
Further information in the text “Data Documentation”
Rights to or over data can be defined from two perspectives. From the researchers’ perspective, these are decision-making powers over the data that arise from the creation of the data. From the users’ perspective, these are the rights that must be observed when reusing data. Rights can be established and communicated in a legally binding form through licences and associated licence texts, as well as agreements.
For the re-use of data, the rules of Good Research Practice apply as a minimum, i.e. essentially the obligation to cite authors correctly (copyright). By granting the Creative Commons licence CC-BY, the data producer can largely replicate this rule under licence law. Restrictions under data protection, patent law and personality rights law may complicate re-use.
A replication study is a scientific investigation that verifies whether the results of a previous study are valid and reliable.
In research, the term replicability refers to the process of repeating a study with new data, whilst retaining the original research design and analysis to achieve similar results. Replicability is central for validating research findings and the credibility of research, as it uncovers errors and discrepancies in the original research and contributes to the integrity of research.
Reproducibility refers to the ability to achieve the same result with the same data and methods. This is important for verifying or understanding research findings. A prerequisite for reproducible results is thorough documentation that adequately describes both the initial conditions and the analysis, as well as access to (raw) data, tools and results.
The term repository refers to a place for storing and publishing digital objects. Generally, a distinction is made between repositories for text publications and repositories for research data – however, there are also hybrid repositories that accept both. Repositories are further divided into generic, discipline-specific and institutional repositories: whilst generic repositories are interdisciplinary in nature, discipline-specific repositories focus on more narrowly defined data types or text publications; institutional repositories can only be used by staff members of the respective institutions or consortia of institutions to publish e.g. data or texts. Repositories are not only suitable for storing data, but also serve to ensure that it can be found and reused by third parties; in the case of research data repositories, this is achieved by making either the data itself or its metadata available. Repositories therefore play a key role when handling research data, as they support the implementation of the FAIR principles. In particular, this is achieved through the provision of metadata, persistent identifiers and standardised interfaces.
Read the article
ROR stands for Research Organisation Registry and is a persistent identifier for research organisations and funding bodies.
It is supported by leading figures from the Open Science community (according to the information provided on its website, ROR is funded, operated and managed by the California Digital Library, CrossRef and DataCite).
S
When creating an object or project, the personality of the creator may be expressed in the result. The extent of this defines the threshold of originality (also known as the level of intellectual creativity), which is decisive in many legal systems for determining eligibility for copyright protection. In Germany, the key factor is that the result is shaped individually by the person, rather than by external constraints (objectives, functionality, objectivity, etc.). If a person’s discernible creation achieves the necessary level of originality, it is referred to as a work.
Works are generally protected by copyright for up to 70 years after the author’s death. This may be relevant, for example, for research data from corpus linguistics or discourse analysis, where contemporary texts may constitute research data but may not necessarily be freely shared. However, the same may also apply to complex questionnaires, which sometimes meet the threshold of originality.
In Liechtenstein, Austria and Switzerland, the threshold of originality is irrelevant; there, it is only necessary that the object or project in question is an intellectual creation and holds a certain degree of individuality (i.e. distinction from what previously existed).
Secondary data is data that has been derived from or processed from primary data. It is thus the result of primary data processing. The use of secondary data can save resources. However, when performing one’s own research or project, critical attention should be paid to how relevant and specific the data is to one’s own research project and how this data may be reused legally.
The term Semantic Web refers to activities aimed at adding a semantic layer to the World Wide Web. This enables the machine-based exchange of information and involves contextualising key terms on a website using metadata, so that it becomes clear, for example, whether ‘Berlin’ refers to the capital of Germany, another city or a name. Machine-readable metadata standards are used to convey the context of a term to machines.
Sensitive data refers to data that, for specific reasons, appears to be particularly worthy of protection. In part, this constitutes a subcategory of personal data and refers to data that contains particularly intimate details regarding a person’s identity. Such data can, for example, be highly valuable for sociological, psychological and medical research, but it is subject to even stronger protection than personal data in the DACH region.
Sensitive research data may, for example, include the following:
- Trade secrets
- Geodata (e.g. on the occurrence of endangered species, such as rare plants; migration routes of animals such as whales or elephants; locations of cultural or archaeological sites; satellite data that may be subject to military use)
- Other information on national security
- “Special categories of personal data” / “sensitive personal data”: Where individuals are concerned, personal data may reveal the following:
- Ethnicity or ethnic origin
- Political opinions
- Trade union membership
- Religious or philosophical beliefs
- Genetic, biometric or health data
- Sexual life or sexual orientation
- Gender identity
Software-as-a-Service (SaaS) refers to software that is usually provided by an IT service provider via a cloud infrastructure against a usage fee, and which the customer typically accesses via a web browser. The main advantage of this business model is that the IT infrastructure required to run the software is provided by the external service provider, who also updates and advances the software. Key disadvantages of SaaS are that the software is only available for a usage fee, which can lead to dependencies. Furthermore, the use of SaaS can be problematic, particularly with regard to sensitive data, as its processing – and often its storage as well – takes place on the external service provider’s servers. Furthermore, the use of SaaS products requires a permanent and stable internet connection.
According to the definition of the DINI/nestor working group on research data (DINI/nestor AG Forschungsdaten), a Software Management Plan (SMP) contains general and technical information about a software project, details on quality assurance, release and public availability, as well as legal and ethical aspects relating to the software.
An SMP summarises information that adequately describes the creation, documentation, storage, versioning, licensing, archiving and/or publication of the software produced or used in a project. Associated hardware and other necessary resources, as well as any related additional software and software libraries as well as text and data publications must also be described and constitute a distinctive SMP feature.
The main purpose of an SMP is to ensure the traceability and, where applicable, the long-term usability of the software (both for direct use and for further processing) and to facilitate user support in the event of queries. An SMP therefore also supports quality assurance (see FAIR4RS Principles).
An SMP may be linked to one or more data management plans (DMPs) if the software is used to generate or process data. SMPs and DMPs can be combined as output plans (see Software Sustainability Institute).
Read the article
T
Terminology refers to the totality of technical terms and names used within a specific discipline or subject area. It serves to unambiguously name technical subjects, concepts and contexts, and to facilitate the most consistent possible understanding within a discipline.
In research data management, terminologies play a particularly important role in describing and documenting research data. The consistent use of technical terms supports the discoverability, comprehensibility and reusability of data and can improve interoperability between datasets and information systems.
This term refers to the application of automated, statistical analysis methods to large datasets. The analysis is usually carried out with computers and complex algorithms in databases. Artificial intelligence (AI) and, increasingly, visualisation techniques are also frequently used in the context of data mining. The aim is to discover previously unknown patterns in the data and to generate hypotheses automatically. The term text mining is used when dealing with unstructured data, particularly in the form of text. Sometimes the term data mining is also used, somewhat imprecisely, as a keyword for the collection, storage and processing of large amounts of data.
A thesaurus is a controlled vocabulary that organises words by subject area. It highlights connections to synonyms, antonyms and thematically related words. It emphasises preferred terms, and lists superordinate and subordinate terms. Significantly more extensive possibilities for describing relationships between terms are offered by ontologies.
U
In continental Europe, the use of literary, artistic and research-related works (i.e. copyright-protected works) is subject to the restrictions of the respective national copyright law. Unless users of these works are granted further rights of use through an additional licence (e.g. a Creative Commons licence), re-use of the works is only possible within the framework of the restrictive permission regulations of the relevant law.
Whether research data is protected by copyright or not depends on whether it meets the threshold of originality. The exact requirements vary from country to country and may be supplemented by related rights: for example, in Germany, datasets not covered by copyright may still be subject to related rights such as the database protection law. As the fulfilment of the requirements must be assessed on a case-by-case basis, it is advisable to consult a specialist lawyer in cases of doubt.
To ensure maximum reusability of research data, which in principle may be subject to copyright law, the granting of additional rights of use, e.g. through appropriate licensing of the data, should be considered. The granting of such licences usually leads to a more ample use of the data in research and can thus contribute to enhancing the reputation of researchers, even beyond the boundaries of the respective specialist community.
Read the article
URN is the abbreviation of an identification and addressing system. Similar to a DOI, it is used for the persistent identification of digital objects (e.g., text publications or datasets). It is particularly widespread in German-speaking countries because the URN service of the German National Library uses, administers, and resolves URNs for persistent identification and addressing.
V
Given that data changes as you work with it, it is recommended to label individual work stages and associated changes via versioning. It is recommended to use a clearly defined, easy to understand versioning scheme that has been agreed upon in advance with other stakeholders (e.g. Version 1.3 or Version 2.1.4). Versioning should take place during the research process itself, for example to mark different working versions of data, and when changes are made to already published research datasets. This provides guidance to those reusing the data and enables them to cite the data correctly. Data versioning can be carried out either manually or with versioning software such as Git.
Virtual research environments (VREs) are software solutions or platforms that facilitate location-independent collaborations between researchers. Most VREs are application-oriented services provided by institutional infrastructure organisations for specific research consortia or communities. Typically, VREs combine subject-specific tools, tool collections, and working environments.
W
Warm Data is research data that is accessed less frequently than Hot Data. This entails that longer access times are acceptable as, unlike with Hot Data, this does not have a negative impact on the work process. This type of data is rarely modified and therefore in a more stable state than Hot Data. In many cases, warm data is already suitable for sharing within a working group or with external researchers, as it has reached a higher level of maturity; for example, it has been enriched with metadata and indexed. Conventional hard disk drives (HDDs), network storage such as SAN (Storage Area Network) and NAS (Network Attached Storage), or cloud-based storage solutions are used to store warm data. These technologies offer a good balance between storage costs and data availability and are therefore particularly suitable for data that is in a transitional phase between active processing and long-term archiving.
X
XML is a markup language used for storing hierarchically structured information as simple text files. This language is mainly used for platform-independent data exchange between applications or computers. The encoding is both machine- and human-readable – this allows the structure and content of encoded information to be described very precisely. With the help of XSL (XML Stylesheet Language), it is possible to interpret the stored information and convert it into other file formats.
Y
YAML (“YAML Ain’t Markup Language”, originally “Yet Another Markup Language”) is a text-based file format used for data serialisation. In research data management, it is frequently used for two main purposes: to create configuration files and to describe metadata. Thanks to its high human-related readability, YAML is particularly well suited to defining software pipelines and data processing steps in configuration files. Compared to JSON, YAML does not use extensive structural elements such as brackets and quotation marks, which makes the syntax clearer. However, these advantages come at the cost of certain compromises.
Although YAML was developed with the aim of simplifying the syntax, it comes with some drawbacks. The indentation-based syntax makes the format prone to errors – even a single misplaced space can cause problems. Furthermore, values without quotation marks can easily be misinterpreted (for example, 04:30 is read as a time rather than a plain text string). Beyond that, there are security concerns, as many YAML parsers allow arbitrary code to be executed by default. Furthermore, different implementations vary, which can limit portability across programmes. For these reasons, it is advisable to always enclose text strings in quotation marks and to use only secure parser functions.
Despite these limitations, YAML’s simplicity makes it suitable for manageable, human-readable metadata descriptions, as via YAML research datasets can be described in a way that is both machine- and human-readable. However, when working with large datasets, under strict validation requirements or if there are high security concerns, JSON or specialised metadata formats should be considered as alternatives.
Z
In research data management, certification generally refers to the formal recognition of repositories that meet certain standards. This makes it apparent to external parties that a repository is trustworthy and of high quality – at the same time, the absence of a certificate does not automatically mean that a repository is of inferior quality.
During the certification process, an independent body assesses, on the basis of various criteria (e.g. organisational and technical measures), whether the long-term curation and provision of data as well as the preservation of its usability, are ensured. A certificate is only awarded for a limited period (for example, three years), after which the repository must undergo a re-assessment.
One internationally recognised certificate for research data repositories is the CoreTrustSeal (CTS).
Depending on the subject area and research discipline, there are different citation practices or guidelines for citing research data publications. There is currently no uniform standard for citing research data, although the Data Citation Synthesis Working Group has formulated interdisciplinary recommendations in its Joint Declaration of Data Citation Principles. One key factor identified in the declaration is the inclusion of persistent identifiers in citations.
Read the article