RECCOMENDATIONS FOR
Linked Open Data

1. Reuse existing resources, such as ontologies, taxonomies, thesauri, tools etc., clearly specifying them in the documentation. #

Linked data are another key building block of the semantic web and are empowered by the same technologies outlined in the ontologies section, in particular RDF, SPARQL, JSON-LD, OWL, and SKOS. Another common specification is RDF-a (RDF in Attributes), which allows to use attributes within HTML5, XHTML and XML documents to express structured data.

• Explore existing linked open datasets and other resources using the Linked Open Data Cloud, the LOV (Linked Open Vocabularies) database, BARTOC (Basic Registry of Thesauri, Ontologies and Classifications), EU vocabularies or the H-SeTIS (Heritage – Semantic Tools and Interoperability Survey) database specific for the heritage domain.
• Domain-relevant vocabularies: the Getty Vocabularies, CLARIN Concept Registry, DARIAH Vocabs services.
• Domain-relevant models and ontologies, used on a national and international level: CIDOC-CRM, FOAF, schema.org, Dublin Core, LRMOO, DataCite, DCAT, SPAR Ontologies, and ArCO.

2. Formalise and release the data model as an ontology, in case it comprises new classes and properties that may be useful for other scholars. #

In this case, follow the recommendations of the Ontologies section.

3. Link your data extensively to external resources such as authority records and vocabularies to reconcile entities and harness the semantic web’s potential. #

In the 5-star guideline, linking data to other data is the fifth necessary step to create excellent open data:

1. “make your stuff available on the Web (whatever format) under an open licence
2. make it available as structured data (e.g., Excel instead of image scan of a table)
3. make it available in a non-proprietary open format (e.g., CSV instead of Excel)
4. use URIs to denote things, so that people can point at your stuff
5. link your data to other data to provide context”

• Check records in existing wide-spread resources such as: Wikidata, VIAF, Open Library, and WorldCat.

4. Ensure data provenance, storing provenance information along with content data in order to prevent inconsistencies when integrating sources, emphasise content responsibility, and foster data credibility. #

• Provenance data can be modeled according to the PROV ontology.

5. Provide the data model and URI pattern to facilitate data reuse. #

• Create persistent URIs, using services such as W3ID.

6. Publish your data in trustworthy repositories in different standard formats and register your linked open dataset in LOD-specific resources like the Linked Open Data Cloud or other domain-relevant repositories. #

7. Offer a SPARQL endpoint with accompanying search examples. Consider implementing a user-friendly GUI for data visualisation and navigation, catering to users unfamiliar with SPARQL. #

While a SPARQL endpoint is the preferred tool for exploring a linked open dataset, a graphical user interface with browsing and searching functionalities can help a wider audience access and use your data.

• For publishing your ontology, you can use visualisation tools such as Widoco and LODE. List of available tools.

List of tools that can be used to create a Triple Store, i.e., an RDF-based database.