Multilingual Linked Open Data Patterns

Linguistic Metadata


Add linguistic metadata, like localization information or the default language of the dataset.


Some linked data applications, like thesaurus, controlled vocabularies, etc. need to have a finer control on the linguistic terms that they are handling. They may need, for example, to express the linguistic relationship between two concepts or the language in which they are expressed.

Given that it is not possible to have literals as subjects in the RDF model, it is necessary to employ resources as literal representatives and to assert declarations between those resources.


The following example shows how we can declare that CatedrĂ¡tico means Professor and that it is a Spanish term (represented by the code spa in ISO-693-3.

    lexvo:means wordnet:Professor ;
      <> .


This pattern exposes the semantic relationships between multilingual labels, so they can be connected with other resources.

Using the property lexvo:language it is also possible to declare the language of a dataset, so one can use the Labels without language tag pattern. However, this option is not a standard practice and users may not be aware of those global declarations.

Although it has been proposed to add a property to declare the default language of a named graph in RDF 1.1, it was rejected.

In JSON-LD, it is possible to declare the default language of a given context.

See also

The Lexvo project defines an ontology of linguistic terms. Lexvo proposes a general framework to publish multilingual knowledge bases. For example, it declares a property language and contains URIs for the different languages.

In this way, language declarations can be part of the RDF model as well as the relationships between terms.