Publishing Linked Data - Linked Data: A Geographic Perspective - page 126

partition (that uses void:propertyPartition ) contains all the triples that use
that property as the predicate.
The final type of metadata that can be described using a VoID vocabulary describes
the links between two sets of RDF triples. Every set of links (a void:linkset )
must have exactly two distinct void:targets (a subject and an object dataset).
For example, if Merea Maps wanted to specify a linkset connecting the Merea Maps
administrative regions dataset to the Merean Mail postcode dataset, it could specify
a linkset as follows:

:MereaMaps_MereanMail a void:Linkset;
void:target http://data.mereamaps.gov.me ;
void:target http://postcodes.mm.gov.me ;

There are several tools for helping you generate a VoID description, including ve2,
the VoID editor, 29 and voidGen. 30 VoID can also be produced within the OpenLink
Virtuoso triple store, 31 which is discussed further in this chapter. The VoID browser
is an easy way to view and query VoID descriptions of datasets. There are several
different options for publishing a VoID description. One is to embed the VoID as
RDFa markup into an HTML page about the dataset, using a local “Hash URI” for
the dataset, for example, http://mereamaps.gov.me/#AdministrativeGeography . An
alternative option is to use a “Slash URI” and along with the 303 redirect mechanism
explained previously in this chapter, to serve the VoID in the same way as the main
RDF dataset. The final option is to encode the VoID description in the RDF format
Turtle. The file is, by convention, called void.ttl, and should be placed in the root
directory of the site with a local “Hash URI.” The main RDF dataset will then have a
URI of the form http://mereamaps.gov.me/void.ttl#AdministrativeGeography .

7.7 PROVENANCE

To facilitate future reuse, it is important to specify where the data has come from:
its provenance. Provenance is defined as “an explicit representation of the origin of
data” 32 and can include information about how a dataset was produced or what facts
were relied on for a decision. On one level, a dereferenceable URI does this implic-
itly by providing information directly from the owner of the URI. However, as Merea
Maps takes data from the Merean Post Office to create its dataset of addresses, it also
wants to be able to specify which fragments of its address data came from this third
party. This provenance metadata can be represented as RDF triples, where the object
of the triple specifies the document in which the original data sits. As mentioned,
the Dublin Core vocabulary can be used to indicate who created and published the
data and when using the predicates dc:creator , dc:publisher , and dc:date .
If the objects of these creator and publisher triples are the URIs of the relevant
people or organizations, rather than just their names as strings, further information
about the provenance of the data can be obtained by querying the dataset for any
triples with the creator or publisher URI as the subject.