JudaicaLink in production

JudaicaLink started as a research project. It is now a service that libraries and researchers can rely on, operated in the portal of the Specialized Information Service Jewish Studies (FID Jüdische Studien) at the University Library Johann Christian Senckenberg in Frankfurt am Main.

Why the move

A knowledge graph is only useful if it is still there in ten years. Research groups change, projects end, staff move on. Libraries, on the other hand, are built for exactly this kind of continuity: they keep collections, catalogues and authority data available across decades.

JudaicaLink was developed at Mannheim Technical University, previously at Stuttgart Media University. Running the production service there would have tied it to the lifetime of a research group. Moving it into the FID portal puts it where the rest of the subject infrastructure already lives, next to the subject catalogue, the repository and the authority data work of the FID.

What changed technically

The move was not a copy. The service was rebuilt to fit the portal:

• JudaicaLink is integrated as a Django application in the FID portal rather than running as a separate system.
• The search backend was migrated from ElasticSearch to Apache SOLR, all indices were rebuilt, and data structures and source code were adapted.
• Operation happens in the Kubernetes infrastructure of the library, with data kept on its central PostgreSQL server — the same environment that carries the portal itself and the Research Navigator. JudaicaLink no longer needs infrastructure of its own.
• data.judaicalink.org now resolves to Frankfurt. Entity URIs keep working and serve HTML, RDF/XML, Turtle or JSON-LD through content negotiation, so references stored in other applications remain valid.
• Dumps and the BEACON file are delivered from Frankfurt. The German National Library receives its updates for EntityFacts from there.

Who does what now

Frankfurt runs the service. Operation, data curation, the dataset catalogue and user support are with the FID. The dataset documentation is maintained there, and that is where additions and corrections belong. For questions about the content of JudaicaLink or about the service itself, the FID team in Frankfurt are the people to ask.

Mannheim develops. This is where work continues on the knowledge graph itself: the data model and ontology, provenance of individual statements using RDF-star, the planned integration with Wikidata, and the processes that turn heterogeneous source data into Linked Open Data. What proves itself here moves into the portal.

That division is the point of the arrangement. Development can take risks; a production service should not.

The older systems

Three systems were in use during development. Only this website remains in Mannheim:

Address	What it was	Status
web.judaicalink.org	this website, the project site	active
data.judaicalink.org	Linked Data delivery	migrated to Frankfurt
labs.judaicalink.org	the prototype that became the production system	being retired

If an old link brought you here, the Labs page lists where each function can be found today.