Using Linked Data - Linked Data: A Geographic Perspective - page 155

SPARQL 1.1, which explains how Named Graphs can be used in SPARQL Update 16
or the REST alternative, the Uniform HTTP Protocol for Managing RDF Graphs. 17
Let us think about how Merea Maps could use Named Graphs to store their
context, for example, the triple stating that the Isis Tavern is near the River Isis:
@prefix mm: < http://mereamaps.gov.me/topo/ >.
mm:isis_tavern mm:near mm:river_isis.
Merea Maps decides to store this triple in a separate named graph so that it can
explain its context (in this case, what is meant by “near,” but the context could
equally explain when the fact is valid). Merea Maps has used the “near” predicate to
state that Medina is near Merea City elsewhere in its data, which is obviously apply-
ing “near” in a different context. Although there are several ways to encode named
graphs, such as the TriG syntax for Named Graphs (Bizer and Cyganiak, 2007),
which is an extension of Turtle, the only way that is self-contained in RDF is to use
the SPARQL 1.1 service description.
In this way, as Merea Maps provides data via its SPARQL endpoint, it includes
a description about the context in which its triple is valid (and whatever additional
information Merea Maps chooses to add in here, such as copyright notices). Note
that the triple has been loaded into the triple store under the specific named graph
< http://mereamaps.gov.me/topo/graph/0012_near > :
@prefix sd: < http://www.w3.org/ns/sparql-service-description# > .
@prefix mm: < http://mereamaps.gov.me/topo/ >.
mm:topo_endpoint
a sd:Service;
sd:url < http://mereamaps.gov.me/sparql > ;
sd:availableGraphDescriptions [
a sd:GraphCollection;
sd:namedGraph mm:isis_near_graph;
];.
mm:isis_near_graph
a sd:NamedGraph;
sd:name < http://mereamaps.gov.me/topo/graph/0012_near > ;
mm:has_context 'Small scale';
mm:copyright 'Merea Maps';.
Note that for many resources, however, it is not necessary to be so picky, and
mostly, Merea Maps assigns a named graph on a per resource basis, so that the con-
cise bounded description (i.e., all its literal properties, related blank nodes, and prop-
erties that link to related resources; see Section 7.6.1) of the resource is contained in
one named graph.
All these techniques rely on being able to explicitly specify the context of use.
However, context is often an emergent property, as shown in Figure 8.1 , where
URIs A, B, and C are tightly coupled (A links to B and vice versa), whereas D links
to A, but A does not link back to D. This makes A, B, and C more of an emergent
cluster than D, indicating that the links between A, B, and C are based on the same