Namespaces and Bindings

RDFLib provides several short-cuts to working with many URIs in the same namespace.

The rdflib.namespace defines the rdflib.namespace.Namespace class which lets you easily create URIs in a namespace:

    from rdflib import Namespace

    EX = Namespace("")
    EX.Person  # a Python attribute for EX. This example is equivalent to rdflib.term.URIRef("")

# use dict notation for things that are not valid Python identifiers, e.g.:
    n['first%20name']  # as rdflib.term.URIRef("")

These two styles of namespace creation - object attribute and dict - are equivalent and are made available just to allow for valid RDF namespaces and URIs that are not valid Python identifiers. This isn’t just for syntactic things like spaces, as per the example of first%20name above, but also for Python reserved words like class or while, so for the URI, create it with EX['class'], not EX.class.

Common Namespaces

The namespace module defines many common namespaces such as RDF, RDFS, OWL, FOAF, SKOS, PROF, etc. The list of the namespaces provided grows with user contributions to RDFLib.

These Namespaces, and any others that users define, can also be associated with prefixes using the rdflib.namespace.NamespaceManager, e.g. using foaf for

Each RDFLib graph has a namespace_manager that keeps a list of namespace to prefix mappings. The namespace manager is populated when reading in RDF, and these prefixes are used when serialising RDF, or when parsing SPARQL queries. Prefixes can be bound with the rdflib.graph.Graph.bind() method:

from rdflib import Graph, Namespace
from rdflib.namespace import FOAF

EX = Namespace("")

g = Graph()
g.bind("foaf", FOAF)  # bind an RDFLib-provided namespace to a prefix
g.bind("ex", EX)      # bind a user-declared namespace to a prefix

The rdflib.graph.Graph.bind() method is actually supplied by the rdflib.namespace.NamespaceManager class - see next.


Each RDFLib graph comes with a rdflib.namespace.NamespaceManager instance in the namespace_manager field; you can use the bind() method of this instance to bind a prefix to a namespace URI, as above, however note that the NamespaceManager automatically performs some bindings according to a selected strategy.

Namespace binding strategies are indicated with the bind_namespaces input parameter to NamespaceManager instances and may be set via Graph also:

from rdflib import Graph
from rdflib.namespace import NamespaceManager

g = Graph(bind_namespaces="rdflib")  # bind via Graph

g2 = Graph()
nm = NamespaceManager(g2, bind_namespaces="rdflib")  # bind via NamespaceManager

Valid strategies are:

  • core:
    • binds several core RDF prefixes only

    • owl, rdf, rdfs, xsd, xml from the NAMESPACE_PREFIXES_CORE object

    • this is default

  • rdflib:
    • binds all the namespaces shipped with RDFLib as DefinedNamespace instances

    • all the core namespaces and all the following: brick, csvw, dc, dcat

    • dcmitype, dcterms, dcam, doap, foaf, geo, odrl, org, prof, prov, qb, sdo

    • sh, skos, sosa, ssn, time, vann, void

    • see the NAMESPACE_PREFIXES_RDFLIB object in rdflib.namespace for up-to-date list

  • none:
    • binds no namespaces to prefixes

    • note this is NOT default behaviour

  • cc:
    • using prefix bindings from which is a online prefixes database

    • not implemented yet - this is aspirational


Note that regardless of the strategy employed, prefixes for namespaces can be overwritten with users preferred prefixes, for example:

from rdflib import Graph
from rdflib.namespace import GEO  # imports GeoSPARQL's namespace

g = Graph(bind_namespaces="rdflib")  # binds GeoSPARQL's namespace to prefix 'geo'

g.bind('geosp', GEO, override=True)

NamespaceManager also has a method to normalize a given url:

from rdflib.namespace import NamespaceManager

nm = NamespaceManager(Graph())

For simple output, or simple serialisation, you often want a nice readable representation of a term. All RDFLib terms have a .n3() method, which will return a suitable N3 format and into which you can supply a NamespaceManager instance to provide prefixes, i.e. .n3(namespace_manager=some_nm):

>>> from rdflib import Graph, URIRef, Literal, BNode
>>> from rdflib.namespace import FOAF, NamespaceManager

>>> person = URIRef("")
>>> person.n3()

>>> g = Graph()
>>> g.bind("foaf", FOAF)

>>> person.n3(g.namespace_manager)

>>> l = Literal(2)
>>> l.n3()

>>> l.n3(NamespaceManager(Graph(), bind_namespaces="core"))

The namespace manage also has a useful method compute_qname g.namespace_manager.compute_qname(x) (or just g.compute_qname(x)) which takes a URI and decomposes it into the parts:

                ("ns2", URIRef("http://foo/bar#"), "baz"))

Namespaces in SPARQL Queries

The initNs argument supplied to query() is a dictionary of namespaces to be expanded in the query string. If you pass no initNs argument, the namespaces registered with the graphs namespace_manager are used:

from rdflib.namespace import FOAF
graph.query('SELECT * WHERE { ?p a foaf:Person }', initNs={'foaf': FOAF})

In order to use an empty prefix (e.g. ?a :knows ?b), use a PREFIX directive with no prefix in the SPARQL query to set a default namespace: