Lexvo.org Frequently Asked Questions

What is the Semantic Web? What are URIs?

Please refer to Wikipedia for more information.

Are the URIs Dereferenceable?

Yes, clients can receive either a web page, or if requested, an RDF data representation of pertinent information about an entity.

Should I use `http://www.lexvo.org/id/...` or `http://www.lexvo.org/page/...`?

You should use the URIs starting with http://www.lexvo.org/id/ to refer to the language-related entities. The URIs starting with http://www.lexvo.org/page/ instead refer to web pages about those entities. If a web browser accesses an URI starting with http://www.lexvo.org/id/, it will automatically be redirected to the corresponding web page.

Why Term URIs?

String literals cannot serve as subjects of an RDF triple. For expressing lexical knowledge, several ontologies have defined OWL classes that represent words or other terms in a language. However, the URIs for individual terms are often created on an ad hoc basis. For instance, the W3C draft RDF/OWL Representation of WordNet has defined URIs for the words covered by the WordNet lexical database.

Linking to term URIs is especially useful to establish the meaning of a non-information resource URI more clearly. For, example, we might have an URI such as <http://www.some.org/#Frankfurt> that is supposed to refer to the city of Frankfurt in Germany. However, we should rely on factual data rather than mere appearances to derive this meaning, because it shouldn't matter to us whether the URI is named <http://www.some.org/#Frankfurt> or <http://www.some.org/#City348914>. One way of doing so is to clarify the meaning using a lexicalization relation:
<http://www.some.org/#Frankfurt> <lexvo:lexicalization> <lexvo:term/deu/Frankfurt%20am%20Main>
or
<http://www.some.org/#City348914> <lexvo:lexicalization> <lexvo:term/deu/Frankfurt%20am%20Main>
Now, it is clear in both cases that the URI can only denote entities that are called "Frankfurt am Main" in German.

Why Language-Specific Term URIs?

Different levels of abstractions can be chosen. We made a pragmatic choice to consider two term entities distinct if the strings are different after Unicode NFC normalization, or if the ISO 639-3 codes differ, which is similar to the RDF semantics for literals. Thus we do not distinguish the meanings of polysemous words in a language, e.g. the verb and noun meanings of the English term "call". In contrast, we do consider the Spanish term "con", which means "with", distinct from the French term "con", which means "idiot".

Why not `<rdfs:label>` or `<skos:prefLabel>` instead of `<lexvo:lexicalization>`?

<lexvo:lexicalization> represents the semantic relation that holds between an entity and terms (words, names, etc.) commonly used to refer to it, e.g. between Albert Einstein and the string "Albert Einstein", or between the concept of books and the French term "livre" (NB: It is deliberately underspecified to apply to real-world entities as well as conceptual entities). RDF triples involving <lexvo:lexicalization> describe actual language use.

In contrast, <rdfs:label> is merely an annotation property that is used to assign human-readable resource labels to resources, which can also be identifier strings such as minCardinality rather than genuine words or names used by a language community.

The SKOS label properties force us to make normative judgments about which lexicalization is preferred for a given entity. This makes sense within a single authoritative thesaurus, but is not appropriate for an open environment where we merely wish to describe which terms are commonly used to refer to something.

Return to main page