SPCData: The Italian Public Administration Data Cloud - Information and Communication Technologies in Public Administration - page 263

of the possible connections among the selected concepts. Every connection was
named with a property that describes the type of connection between two dif-
ferent concepts. Afterward, we evaluated all the well-known and wide-accepted
ontologies and vocabularies to find properties and concepts corresponding to ours.
The reuse of common ontologies and vocabularies has the merit of making our
data set easier to understand for practitioners and machines and of guaranteeing
semantic interoperability with the web of data. For instance, we used the ontology
Friend of a Friend (FOAF; 2012) (i.e., the property “ homepage ” to denote all the
web site URLs of Italian Pas), the ORG ontology to represent organizations, and
Data Catalog to represent catalogues of data sets. In addition, it makes large use of
the core vocabularies, recently promoted by W3C and the European Commission.
Some of the property we specified in the data set has been directly taken from
the standards Ontology Web Language (OWL) (W3C, 2004), RDF, and RDF
Schema (W3C RDF Vocabulary Description Language 1.0: RDF Schema, 2012).
For instance, we represent the description of the services ( des_srv in the previ-
ous IPA data example) with the existing property rdfs:com ment .

12.4. 3 Mapping

Once both previous phases have been completed, the RDF triples reflecting the
given IPA conceptual model have to be produced. In doing so, we carried out an
assessment of current reference documentation (Sahoo et al., 2009) of the W3C
RDB2RDF Working Group to survey all the existing approaches that can be applied
at this stage. Among them, we opted for D2RQ platform (Bizer & Cyganiak, 2007;
Freie Universitat Berlin, Web-based Systems Group, 2012), which is a comprehen-
sive and flexible framework for our purposes. D2RQ includes, among others, a
declarative mapping language that allows one to transform data included in a rela-
tional database into RDF triples. We implemented a specific D2RQ script that
helped us to automate the whole mapping process.
Figure 12.3 shows an example of a D2RQ script clause we used to map the
previously illustrated IPA records into RDF triples.
The type of clause ( d2rq:PropertyBridge ) illustrated in this figure serves for
connecting literals (see d2rq:colu m n ) to entities that belong to some concept (see
d2rq:belongsToClassMap ) through a specified property ( d2rq:property ).
In our example, we connect web site URLs (contained in db _ servizio _
amministrazione.uri _ srv ) of PAs' services (ServizioAmministrazione

Figure 12.3 D2RQ script clauses used in the mapping phase.