Using Linked Data - Linked Data: A Geographic Perspective - page 150

8.5.1 S tep 1: S pecify the p urpoSe , S cope , and c oMpetency Q ueStionS
Linking datasets can be treated in the same way as authoring a new dataset, follow-
ing the steps set out in Chapter 7: First, the purpose and scope of the link ontology
must be specified, and a set of competency questions can be developed (which can be
expressed in SPARQL) for testing the links. Again, using Merea Maps as our exam-
ple, Merea Maps' link architect wants to link the administrative regions' Linked
Data set that they have just published to other datasets. Merea Maps could have two
quite distinct purposes for this linking: One is to improve discovery , in which case
identity and similarity links should be created, while the other would be for repur-
posing if Merea Maps had found some other data sources that it wanted to pull into
its own dataset, thus enlarging the set of data available to its customers in a specific
application, in which case relationship and vocabulary links would be required. So,
the Discovery Purpose could be stated as: “To describe correspondences between
Merea Maps' administrative regions and Merean Mail addresses to enable users to
identify the administrative regions in which addresses lie.” In contrast, the purpose
of links created for a Repurposing use case could be stated as: “To describe Merean
Mail addresses in terms of Merea Maps' vocabulary.” This would be needed if, for
example, we wanted to use Merean Mail data within an application that only handled
Merea Maps' administrative data. It should be noted that in many cases both discov-
ery and repurposing will be desirable, but it is still useful to recognize the differ-
ences and explicitly state these purposes.
The purpose is, in effect, stating the context in which the links are valid, and
the scope provides the coverage . The scope of our first Discovery use case could
be stated as, “to cover the description of addresses within Merean administrative
regions.” Classes that fall outside the intersection between the two datasets, such
as Post Office Box Numbers, need not be modeled in the linking ontology. In the
Repurposing use case, the scope of the Linked Data sets would again be the inter-
section of the Merea Maps and Merean Mail datasets but limited only to terminology
used by Merea Maps.
8.5.2 S tep 2: i identify d ata S ourceS
As with the creation of a stand-alone ontology, the second step in Link Ontology
creation is to examine the two datasets and two ontologies describing those datasets,
also taking into consideration other vocabularies that may provide useful relation-
ships for reuse. Let us take the Discovery use case mentioned: Merea Maps wants to
link to Merean Mail data to make its own data more easily accessible on the Linked
Data Web. In this case, Merea Maps needs to look at its administrative regions'
Linked Data, which was extracted from their database of Table 7.1 and modeled as
Linked Data in Chapter 7. Classes and example instances are shown in Table 8.1 ,
including a more detailed explanation of the terms used, to help us when we need to
align them with the Merean Mail addresses.
Merea Maps then compares its own Linked Data against the Merean Mail Linked
Data, as shown in Table 8.2 .