The Labeling System project is developing a prototype web app for creating and publishing terms with contextual validity as Linked Open Data. The system is based on Linked Data concepts, technologies and formats. The core is the modelling of semantic relations and meanings of terms to authoritative HTTP-addressable thesauri.
The indexing of cultural-historical materials requires abstraction into keywords. The contextual compilation of these terms in the form of a vocabulary is a component of many digital research projects. The identifier of a term is not sufficient for an unambiguous definition. A concretisation can be done by descriptive text or linking to external resources on the web. Especially for interoperability between distributed databases, unique identifiers (URIs) are suitable for automated evaluation of relationships. The Linked Data Cloud provides machine-readable authoritative HTTP resources of different scientific disciplines (e.g. Getty AAT) for linking. These are based on the Resource Description Framework (RDF), so publishing the terms as Linked Data makes sense.
A common understanding of the meaning of various context-bound terms within humanities research disciplines is a central challenge, especially in the field of “Digital Humanities”. To this end, these terms must be published and clearly described. Virtual libraries such as Europeana.eu rely on unambiguous indexing using RDF.
The generation of linked terms as linked data is a complex process that is difficult for the humanities to implement with existing frameworks. The goal is therefore a system that enables digital humanists to create semantically related terms and publish them as Linked Data. This system is centrally hosted by an institution (e.g. mainzed) and enables the associated scientists to exchange interdisciplinary information without installing a server. The long-term goal is to use the LS as an infrastructure component, a term gazetteer or meta-index in the “mainzed network”.
The prototype of the labelling system offers scholars the possibility to create terms with context-bound validity, to concretise them, to group them in containers (vocabularies) and to share them with the research community. It provides user-friendly web-based tools that make it possible to semantically link one’s own terms with the Linked Open Data Cloud. The labelling system serves as a decentralised term repository that provides citable addresses as URIs on the web.
Matthias Dufner and Axel Kunz developed a user-friendly frontend based on AngularJS, HTML5 and CSS3 in 2016. The backend implemented by Florian Thiery is based on JAVA, an RDF4J Triplestore and a documented API. The modelling of vocabularies and terms is implemented by a simplified SKOS model adapted to humanities issues.
The cornerstones are the descriptive description of a term (concept) by means of a label, translations and a short description, as well as the hierarchical, associative or mapping linking of external resources.
The GUI enables easy linking of concepts with reference thesauri (e.g. Getty) by keyword search, linking of web resources of the Internet Archive and prototypical work in draft mode.
The competence area WissIT of the RGZM already uses the LS to publish terms (e.g. in the reengineering of the NAVIS databases) and incorporates the terms into its own research. In particular, the use of the LS as a central node of the distributed database architecture (meta-index) is in the foreground here.
In 2017, the prototype is scheduled to be published as one of the central components of the mainzed network’s research data infrastructure, which is currently being set up, at http://labeling.link. This will enable mainzed partners to publish their subject vocabularies and metadata keyword lists centrally as LODs and to integrate them into their system architecture.