AI Ethics Video Series

NFDI4DS develops training materials and organises events that deal with processes and methods of data science and AI.
In this context, a video series on ethical aspects in AI was produced. For this purpose, a curriculum was developed and a questionnaire was created for the series. Each episode highlights a specific aspect and interviews well-known experts in the field. The target audience is practitioners from the data science and AI community.

You can find ll currently published videos of the series on YouTube

Attended Cloud Housing

Data from High-Performance Computing (HPC) and High-Performance Measurements is often stored in personal (project) accounts at data centres and is immobile due to its size. Access to the data or its further use was previously not possible without an explicit request to the data centre, as data on large computer systems is commonly only accessible by the producers and project participants.

With the establishment of a cloud server system at the Leibniz Supercomputing Centre (LRZ) and the Gauss Centre for Supercomputing, a platform has been created that implements the ‘accessibility’ aspect of the FAIR principles in practice. The cloud server system makes it possible for research data generated at the Leibniz Supercomputing Centre (LRZ) to be used by external parties. Access can be customised via a virtual machine or microservice. Thanks to the direct integration into the data storage infrastructure of the data centre, both “hot” and “cold” HPC research data can be accessed. It is also possible to analyse the data on the virtual machine. The huge, immobile data volumes are directly accessible via the cloud and can therefore be reused. The cloud system offers the following utilisation options:
• • Provision of exclusive usage rights within a professionally managed compute cloud.
•• Running virtual machines on the cloud to analyse the data on site
• • Fast connection to existing LRZ data storage systems..

Coscine

Die Forschungsdatenplattform Coscine combines an implementation of the FAIR principles with federated, industrial-grade and scalable data storage systems and provides provisioning and access management processes.

Coscine is being continuously developed in NFDI4Ing to meet the needs of engineering scientists. Coscine.nrw is already offered as a federated service for 42 universities in North Rhine-Westphalia, providing FAIR data management for a broad community of researchers.

In 2023, Coscine successfully transitioned from the pilot phase to regular operation and now serves over 1,500 interdisciplinary users from 138 universities and other research organisations from the Research Organisation Registry. Coscine already supports over 1000 projects in the field of engineering sciences. The platform’s low-threshold access management based on DFN AAI and ORCID enables broad collaboration beyond the academic sector.

As a central service, Coscine promotes collaboration within and between NFDI consortia. Coscine is already being used intensively in NFDI4Ing and NFDI-MatWerk. In NFDI4Chem and NFDI4Microbiota, Coscine is being evaluated in individual working groups, which shows the great potential for cross-network synergies.

Data Collections Explorer

The Data Collections Explorer (DCE) is an easy-to-use information system for the engineering sciences with a low barrier of entry. It provides a quick overview of subject-specific repositories, archives, and databases, as well as of data sets that were published outside this established infrastructure.
Crucial information, such as Open Access policy, API access, publication cost and data set size limits, is clearly represented. This helps scientists that are either looking to publish their research data or are searching for data to advance their research.
Since its launch in March 2022, the Data Collections Explorer has rapidly gained traction in NFDI4Ing, owing its growth mainly to the many contributions by domain scientists not limited solely to the engineering domain. The Data Collections Explorer also garnered interest in other consortia as well as with initiatives outside NFDI. As a first step to broaden the impact of our approach, we are working on expanding the Data Collections Explorer to the materials science and engineering community within NFDI-MatWerk. Additional improvements in the works include setting up a knowledge graph, which will allow an easier integration with existing and upcoming projects amongst other benefits.

Data Steward Service Center

he DSSC of our young consortium was successfully formed within a few weeks of the project start in March 2023 and is characterised by a combination of teamwork and independent task processing. We are five data stewards and one coordinator who deal with various enquiries from the national and international agrosystems community. Each of our data stewards is an expert in a specific area of agrosystems research. This includes geodata, crop inventory data, long-term agricultural trials and genetic data. We are also the first consortium to have a fully qualified lawyer as a data steward for legal and ethical issues. This enables us to answer the community’s questions directly and competently ourselves, so that no outsourcing is necessary. This makes our service agile and efficient. Specifically, enquiries reach us via the central email address “dataservice@fairagro.net”. These are then converted into tickets by our coordinator and assigned to one or more data stewards according to their expertise, who then process and manage these tickets independently. By the end of 2023, we had already dealt with 27 community enquiries. Thanks to the size and diversity of our team, we also successfully contribute to networking with other helpdesks (both NFDI and local) and with the national initiatives. [Author: Lea Singson,Data Steward at DSSC]

The leading institution of the DSSC is the Leibniz Centre for Agricultural Landscape Research (ZALF) with
its own repository (BonaRes Repository).

.

Education Gitlab

The NFDI4Ing Education Gitlab is a platform for the provision of and collaboration on open educational resources (OER) in the field of research data management (RDM). It is an interdisciplinary project based in the NFDI4Ing consortium, but can be used and extended by anyone with a GitLab account. It is a valuable resource for anyone who is interested in RDM or wants to share their knowledge about it.

On the Education Pages, NFDI4Ing offers training courses on research data management that focus on applications in engineering. By using engineering-specific use cases, examples and features, the training courses are specifically tailored to the needs of engineers. Elements such as videos and quizzes have been integrated to make the content interactive. The training courses cover the entire lifecycle of data, covering topics such as data management plans and data reuse, as well as cross-cutting fields such as metadata, file formats, software development and licences.

The training courses and the associated platform have been created with aspects of open educational resources (OERs) in mind: The trainings are open and available without registration and are licensed under CC BY 4.0. In order to use open and established platforms and formats, the trainings are implemented on GitLab pages as platform, markdown and HTML files for the content and H5P for interactive elements. In addition to curation by the NFDI4Ing Training & Education Team, users are invited to share their feedback via the GitLabs Issue System to continuously improve the service.

For more information on RDM training courses for engineering, please click here

.

Unique exchange platform that pools expertise on research data and research data management at a low threshold

Das Forum4MICA – Making Information Commonly Available facilitates the exchange between researchers as data users and research data centres (RDCs) as data providers. In the first few months, 15 of the 41 RDCs have already signed up as partners and almost 400 people have registered as active forum participants, and the trend is rising..

In the forum, researchers can (a) ask their individual questions, (b) contribute their own experiences and suggestions for solutions and (c) systematically search for information. In view of the size and complexity of the datasets in the multidisciplinary field of KonsortSWD, the forum has created a simple, sustainable and low-threshold access route to knowledge about relevant research data infrastructures for the scientific community.

Standardised data in the humanities and cultural sciences

In Text+, the NFDI consortium for language and text-based research data, the Deutsche Nationalbibliothek and the Niedersächsische Staats- und Universitätsbibliothek Göttingen re working together with other partners to expand the indexing of research data in the humanities and cultural sciences with standard data from the Gemeinsamen Normdatei (GND)and to enhance previously undeveloped databases with standard data Standard data refers to research data that can be clearly assigned to a person, for example with the help of an identification number, regardless of factors such as spelling.This benefits research and the public alike, as entities in research data (persons, works, geographies, subject headings, etc.) identified as authority data can be clearly referenced and interoperable, they offer clear search access points for data searches and, when used as controlled vocabulary in different information resources, can drive the development of the Semantic Web forward.

Open Research Knowledge Graph

Academic knowledge. Comparable.

The core idea of NFDI4DS is to increase the transparency, reproducibility, and fairness of data science and AI projects by making all digital artefacts available, networking, and providing innovative tools and services.

The Open Research Knowledge Graph (ORKG) is a key component of this vision. It aims to describe research in a structured way. With the ORKG, work is easier to find and compare.

・ ORKG: https://orkg.org/
・ Stories from researchers: https://orkg.org/about/36/ORKG_Stories
・ Quick overview for beginners (to get started with the Open Research Knowledge Graph and its features): https://orkg.org/about/14/Get_started

Storage space is expensive, and the demand among researchers is high. Most existing systems for allocating storage space are ad hoc, require (internal) transfers of funds, do not scale institutionally or even nationally, and generally cause some formal overhead. In NFDI4Ing, we want to standardise the application and allocation of scientific IT resources and simplify the administrative process.

Based on our existing data management platform Coscine, we have customised Jards, a tool that is already used in many data centres to manage data time requests, so that it can also handle storage requests. Jards then enables a scientifically-led examination of the application and the work and data flows, thereby ensuring a sustainable handling of the research data and metadata. Jards offers an application process that evaluates proposals based on the scientific value and implementation of the FAIR principles. It is available as a federated service for 42 universities in North Rhine-Westphalia and the NFDI consortia in which they participate. Among other things, you use it to coordinate the distribution process for storage resources provided by the DataStorage.nrw consortium.

Technical Chemistry Laboratory

The Faculty of Chemistry at the Technical University of Munich offers the laboratory internship “Technische Chemie” for students of chemistry and chemical engineering. The lab internship includes catalysis-related experiments conducted by the students. Typically, about 100 students are enrolled each semester, and each student performs at least eight different experiments. As a result, more than 800 experiments and reports are produced each year, and are evaluated by supervisors (usually PhD students and postdocs). We have developed a research data management tool, RDM4Lab, to systematically store the data generated in this lab course in compliance with the FAIR principles (Findable, Accessible, Interoperable, and Reusable). The tool also includes features such as visualisation of current and historical data and automatic data analysis to help maintainers evaluate reports.

Legal Helpdesk

A competent contact point: The Legal Helpdesk by NFDI4Culture answers questions regarding copyright and data protection

“What do the changes in copyright law mean for our digital collections?”, “What should we bear in mind for valid consent to data processing?“ As part of the NFDI4Culture Helpdesk the Legal Helpdesk is a service offering competent information to deal with these and similar questions. The spectrum of topics ranges from data protection and copyright to ethical aspects of research data management.
The Legal Helpdesk is open to researchers, cultural heritage specialists, cultural heritage institutions and research projects. It has grown into a reliable and highly demanded service for our community, covering simple questions as well as complex dialogues on data utilisation plans and rights management strategies.

More Success Stories by NFDI4CUlture on the NFDI4Culture Portal!

LARASuite

Die LARAsuite is an open-source, next-generation research data management system: it implements a radically automatable approach so that data and metadata input is reduced to the bare essentials. It uses open-source laboratory communication protocols, open data and metadata standards and a semantic, ontology-based representation of the data.

Data can be exchanged between different LARA instances and synchronised with (inter)national repositories. Queries can be carried out using SPARQL.

LARAsuite and many of its components are developed by an international open source community. It can be used in many scientific fields.

NOMAD

NOMAD is a digital infrastructure for the wide field of material sciences. The consortium FAIRmat has further developed the original collection of simulation results in many directions.This not only concerns calculations of excited states using many-body theories and classical molecular dynamics; NOMAD has also grown into a platform for sample production, various experimental methods and a variety of applications FAIRmat has developed (meta-)data models and graphical interfaces to process and describe the wealth of heterogeneous dataToday, NOMAD comprises 13 million entries for more than 3 million materials.

The interactive NOMAD apps allow the domain-specific data to be explored in detail and also compared with each other. These apps offer customizable user interfaces for data visualization. For example, the Solar-Cell App contains over 40,000 records of components from the Perovskite Database Project. The data collection on Metal-Organic Frameworks, contains more than 17,000 calculations and is currently being expanded to include information on their synthesis. More apps will be available soon, ranging from heterogeneous catalysis to battery research.

With NOMAD Oasis FAIRmat also brings our data infrastructure to individual laboratories. NOMAD Oasis enables researchers from various disciplines to customize the NOMAD software to their individual needs. Visit out website at: https://nomad-lab.eu.

Universitätsvorlesung für mathematisches Forschungsdaten-Management

A survey conducted in the summer of 2021 c German mathematics faculties concluded revealed that teaching mathematicians estimate the awareness and knowledge of their students regarding good scientific practice, authorship attributions, the FAIR principles, and research software as too low. These are classical research-data management (rdm) topics. Motivated by that need and by successful RDM-Courses at the universities in Bielefeld und Leipzig, , six lectures in research-data management for mathematicians took place in Leipzig in the summer term 2023. To the teacher’s knowledge, this was the first of its kind. The large group of attendees came from a variety of career levels including undergraduate students, PhD students, postdocs, and MaRDIans. This contributed to lively discussions centered around properties and common problems of mathematical research data, metadata standards for papers and the difficulties in deciding appropriate metadata for mathematical results, the scientific method, good scientific practice, and how to write, cite, and document mathematics.
.

Feedback for the course was very good, with students appreciating the interactive atmosphere, the time allocated for questions, and the informal nature of the classes. A one-day course of maths rdm in Magdeburg in November built on these first successful sessions and discuss questions of reproducibility and repositories, in addition to introductory topics. Lecture notes for both are now in the making. They will be made publicly available for a second installment next summer term for free use and reuse by any mathematician interested in the topic of rdm.

NFDI Meta Portal 

The key idea of NFDI4DS is to work towards increasing the transparency, reproducibility, and fairness of Data Science and AI projects, by making all digital artifacts available, interlinking them, and offering innovative tools and services.
The key idea of NFDI4DS is to work towards increasing the transparency, reproducibility, and fairness of Data Science and AI projects, by making all digital artifacts available, interlinking them, and offering innovative tools and services.
The Metaportal is based on a data management ecosystem from the public sector

Metadata Shaper

In order to describe a heterogeneous field with metadata, a language is required that can communicate across domains. To make this possible, metadata standards are used as a simple and universal vocabulary to create an interface between knowledge domains. These can include a wide range of research areas, such as catalysis research, chemistry, and process engineering.
A form of such a metadata standard is a metadata framework that not only represents a standardisation of language, but also enables a uniform description of data in a simple but meaningful form. This makes it easier to map the data in knowledge graphs and thus simplifies the creation of searches for these knowledge graphs.

To make data FAIR (accessible, interoperable, reusable) and available for further research, we need to develop a language, a semantic web, that can describe the content of the data in a way that is readable to humans and machines. An essential semantic tool is ontology, which uses a mathematical-logical language and annotation to build an understanding of the data content through references, definitions, and explanations. Since the development and expansion of these ontologies is labour intensive and knowledge intensive, this is typically done only by domain and semantics experts. In NFDI4Cat, we have developed tools to help domain experts develop ontologies using Natural Language Processing (NLP). In the near future, it will be possible to include the increasingly emerging large language models. Suggestions for new ontology content are well supported, along with references to existing content and information extraction based on it.

Ontology World Map

By further developing the research data infrastructure FEI, it is becoming increasingly important to deal with ontologies, which are the logical framework for the creation of knowledge graphs and thus serve the standardised networking of (meta-) data. The NFDI4Cat consortium examined the ontologies that map the themes of catalysis and assessed their usability in the NFDI.
This provides a systematic approach to collect ontology metadata, classify it based on catalysis subdomains, and create a human readable representation in a GitHub repository. It also enables efficient comparison and automated mapping between ontologies, focusing on adaptability to other knowledge domains. The repository is intended to serve as both a focal point for new researchers and a discussion forum for all researchers interested in ontologies (and metadata standards) in the field of catalysis and process-related sciences.

Initial overarching offer for a significant proportion of social science researchers

QualidataNet is a network of research data centres, archives, and repositories focused on qualitative research data. As a “single point of entry”, QualidataNet bundles information about research data and offers of different archiving infrastructures.

Researchers can find data and archiving partners and get information on how to use and managen.

Research data centres, archives and repositories increase the visibility and traceability of their data and network with other providers.

The portal went online at the end of December 2023. The functionality is constantly expanded.

The Research Data Management Organiser by NFDI4Ing manages large amounts of data