NFDI4Culture Metadata API

Terminologies are important links for the interlinking of information, hence they are the subject of the NFDI4Culture working area "Standards, Data Quality, Curation" (Task Area 2). The use of terminologies in the form of online specialist vocabularies, standard data hubs and application ontologies contributes significantly to the implementation of the FAIR Data Principles: They promote and facilitate data exchange across knowledge domains. In digital environments, they are crucial for the findability and interoperability of research data and thus increase their visibility and reach. By categorising content and identifying entities in digital sources, they make a significant contribution to the quality of data.

The aim of the two-day forum on 20/21 March 2024 was to explore overlaps in the understanding of terminologies, in the demands and offers the Text+, NFDI4Memory, NFDI4Objects, NFDI4Culture and Base4NFDI consortia and to provide an overview of services, applications and tools. The consortia presented their current work priorities and initiatives on the topic of terminologies - a good starting point for the subsequent discussions. With a total of 273 registered participants, there proved to be great response.

The introductory talk by Alexander Faschon (NFDI4Culture, SLUB Dresden) on the first day first defined terminologies as terms embedded in conceptual orders, which can be structured in simple to hierarchical ways, and introduced the possibilities of eliminating ambiguity through terminological control. He presented the various forms of controlled vocabularies and their differences, from word list to classification, systematics, taxonomy, thesaurus and standardised data through to ontology.

Barbara Fichtl and Susanne Al-Eryani (both SUB Göttingen) reported for Text+ on the establishment of the GND (Integrated Authority File) agency, which is to take on the tasks of data delivery, quality assurance and editorial work. The agency will primarily be responsible for determining requirements, providing advice and information on the required quality level of the data as well as technical services. entityXML was presented as the agency's internal storage and exchange format, which is intended to offer researchers project-specific data storage for standardised data entities and facilitate data delivery to the GND.

Katrin Moeller (Historic Data Centre Saxony-Anhalt, Martin Luther University Halle-Wittenberg) outlined the plans of NFDI4Memory on the way to a dedicated GND agency for the historical sciences. The needs of the communities were identified in a survey: teaching data literacy for the use of vocabularies in science, dealing with particular challenges, such as the choice of suitable terminology according to subject-specific requirements, especially for items of geography, subject terms and events. For example, it seems appropriate to be able to include terms found in the sources in the vocabulary with supporting documentation. In addition, there is the technical integration of terminology and a need for supporting tools. A large number of specialised vocabularies are used, which will be incorporated into a terminology register to be set up. In addition to the GND as a target system, the option of using platforms such as FactGrid is also available.

For NFDI4Objects, Anja Gerber (Klassik Stiftung Weimar) and Frank von Hagel (Institute for Museum Research, SPK) presented the positions of their consortium in dealing with terminologies. The main focus of interest there is the development of a CIDOC CRM-based data model for object biographies that can reflect the life cycle of objects with details of actors, times, geographies, events, provenances and contexts, including sources and the representation of uncertain knowledge. The archaeological sciences also frequently integrate scientific measurement data from excavations and data from reconstruction and prospection methods. In particular, actors, locations, keywords and time periods are covered by large reference vocabularies. Specialised vocabularies for archaeology and for certain types of objects are also used.

The series of talks continued on the second day with a presentation on NFDI4Culture. Angela Kailus (NFDI4Culture, DDK - Bildarchiv Foto Marburg, Philipps University Marburg) presented the Culture Knowledge Graph and the NFDI4Culture Ontology as well as the vocabularies used in the communities, their fields of application and the most important application scenarios. For practical work with vocabularies, the contribution provided decision-making guidance on the selection, use and technical integration of published terminologies. The integration of vocabulary platforms and services was recommended. The publication of own vocabularies or possibilities for supplementing the large reference vocabularies were also discussed. NFDI4Culture is already supporting its communities in these areas. Other focal points include cooperation with the GND, particularly with regard to authority data for buildings/works and the development of tools to support authorities-based data annotation.

The presentation by Roman Baum (ZB Med - Information Centre for Life Sciences) was dedicated to the TS4NFDI terminology service currently under development. As a basic service for all domains within the NFDI, it offers the possibility of modular, scalable and easily configurable integration of terminologies. Users can search in external vocabularies and compare terms ( assignment of meaning, synonym list). Later expansion stages will include the integration of additional terminologies, including local specialist vocabularies, mapping functions and the option to add new terms. The service supports semantic searches and the annotation of data while at the same time being user-friendly. In a final excursus, Kolja Bailly (NFDI4Culture, TIB Hannover) presented the ANTELOPE annotation and terminology service from NFDI4Culture, which is being developed for typical usage scenarios in this community.

The following key topics emerged from the discussions:

• There is a great interest in using GND data in a variety of research contexts. The desire to collaborate with the GND was often expressed. However, there was a need for clarification with regard to the possibilities of contributing one's own research data to the GND as authority data and the practical implementation (input templates, workflow, permanent maintenance/editing). The participants also requested information on the task and role of the GND agencies and possible permanent operating models.
• The general problem of selecting the most suitable vocabulary for one's own objectives. It should be possible to represent the historicity of terms and the complexity of specialist language, but this stands in the way of standardisation and harmonisation. Specialised vocabularies should therefore be included in the requirement to integrate and interlink different terminologies. This is just as important as an appropriate translation of terms into other languages and of the in-depth description of the meaning and distinctions between terms.
• Relevant for all consortia: the changeability of place names in different eras and the mapping of spatiotemporal categorisation systems. For places, it should be possible to specify differentiated roles and to indicate uncertainty regarding the source situation and localisability. In addition, geo-norm data should be supplemented by the specification of surface coordinates (polygons) or names in other languages and writing systems. A constant challenge is multilingualism and the associated requirement to translate the meaning of specialist terms as precisely as possible. The problem is exacerbated in the context of decolonisation when Western-style terminology is applied to objects and contexts of meaning in non-Western societies.
• Centralised terminology services promise significant support for the integration of vocabulary functionalities in specific and wide-ranging application contexts. In order to achieve greater acceptance, such services should be easy to implement and intuitive to use. Specific use cases can be helpful in customising the offering, as can tutorials and materials for integrating your own vocabularies. Time savings and strategic added value should be emphasised. The persistence of the offering, reusability through open-source publication and the use of permanently available external services (e.g. for AI components) must be guaranteed.

To sum up, the following overlaps in the interests of the participating humanities consortia can be identified: The role of the GND in relation to other vocabulary resources and initiatives is central. The question was also raised as to whether the organisation of GND agencies should be reconsidered in order to be more flexible in response to the needs of the communities represented.

There were also repeated discussions about what vocabularies should and can actually do for the communities. Although the GND covers all subject areas, it is often inadequate for specific research questions. It is therefore important to determine the role of specialised vocabularies in relation to the large reference vocabularies. While specialist vocabularies are created and expanded for very specific documentation and use requirements, the value of the large reference vocabularies lies in ensuring interdisciplinary interoperability. Differentiated, high-quality and comprehensive mappings of published specialist vocabularies to general vocabularies represent the bridge over which data indexed with specialist vocabularies can be more easily found, interoperated and reused. However, this also means that general vocabularies such as the GND must have sufficient coverage of items to enable adequate functional mapping.

It was also pointed out that researchers and those working in cultural heritage collections need more support in acquiring relevant skills, especially in the use of existing LOD vocabularies, in the development, maintenance and publication of their own vocabularies and the addition of external vocabularies, in the referencing of existing databases, the use of tools for the management and integration of terminologies. AI methods promise support, but the results still require subsequent validation by human expertise. However, the future potential can be realised through training with relevant structural data.

The presentations of the lectures, the digital whiteboard, links shared during the event and references are available for download HERE.