IoT Semantic Interoperability for Active and Healthy Ageing
327
To enable semantic interoperability across non-interoperable systems, there are
three different main techniques [24, 25]:
• Ontology alignment: refers to the set of semantic correspondences between
several ontologies. Relations between different ontologies entities are defined in
the alignment. Ontology alignments can be simple (among atomic entities) or have
a higher complexity (across entity groups and entity sub-structures). An alignment
includes predicates of similarity (i.e. matching) such as equivalence and subsumption axioms, or logical axioms (i.e. mapping). Frequently, tools for alignment
definition report the level of accuracy and fiability of each inner correspondence.
• Ontology merging: this technique provides a semantic interoperability solution
across systems by means of the combination of two or more ontologies into a
common one, creating a new ontology that supports the knowledge from the original ontologies. The most simple application of ontology merging would be a
resultant ontology consisting on the sets of axioms from all the original ontologies. The inclusion of new axioms derived from the relation between the source
ontologies would create a more complete and complex global ontology. These
new axioms are typically obtained via ontology alignments.
• Ontology translation: this technique for achieving semantic interoperability
consists on the process of changing the semantics of a piece of information
expressed through an ontology into the semantics of a different ontology. Thus,
information described semantically in terms of a source ontology is transformed
into information described in terms of a target ontology. The result contains information interpretable in the scope of the target ontology semantics. A requirement
for a successful semantic translation is the preservation the original information,
which must not suffer alterations on the meaning. Notionally, no information
must be destroyed through translation. Also, in ideal terms, it should be possible
to revert the translation process. Thus, the original content could be recovered by
performing a reversed translation (from the final ontology into the source one).
Those translations typically require previous ontology alignment.
Additionally, there are several recommendations and good practices for facilitating the achievement of semantic interoperability among systems [23] and reducing
the heterogeneity of information models for a same domain:
• reuse of existing ontologies at the possible extent: reuse of existing knowledge,
avoiding the construction of heterogeneous models that hinder interoperability. It
also implies to use recommended core ontologies in order to create a new ontology
covering the system needs, such as SSN in IoT domains.
• creation of ontologies following best practices [26].
• update and maintenance of ontologies [23].
• creation of catalogs of ontologies for easing the reuse of ontologies, and also their
update and maintenance or the creation of new ontologies aligned with appropriate
and core ones [23, 27].
The accomplishment of these recommendations makes more feasible that systems
follow a common ontology and share common semantics. Also, it significantly
327
To enable semantic interoperability across non-interoperable systems, there are
three different main techniques [24, 25]:
• Ontology alignment: refers to the set of semantic correspondences between
several ontologies. Relations between different ontologies entities are defined in
the alignment. Ontology alignments can be simple (among atomic entities) or have
a higher complexity (across entity groups and entity sub-structures). An alignment
includes predicates of similarity (i.e. matching) such as equivalence and subsumption axioms, or logical axioms (i.e. mapping). Frequently, tools for alignment
definition report the level of accuracy and fiability of each inner correspondence.
• Ontology merging: this technique provides a semantic interoperability solution
across systems by means of the combination of two or more ontologies into a
common one, creating a new ontology that supports the knowledge from the original ontologies. The most simple application of ontology merging would be a
resultant ontology consisting on the sets of axioms from all the original ontologies. The inclusion of new axioms derived from the relation between the source
ontologies would create a more complete and complex global ontology. These
new axioms are typically obtained via ontology alignments.
• Ontology translation: this technique for achieving semantic interoperability
consists on the process of changing the semantics of a piece of information
expressed through an ontology into the semantics of a different ontology. Thus,
information described semantically in terms of a source ontology is transformed
into information described in terms of a target ontology. The result contains information interpretable in the scope of the target ontology semantics. A requirement
for a successful semantic translation is the preservation the original information,
which must not suffer alterations on the meaning. Notionally, no information
must be destroyed through translation. Also, in ideal terms, it should be possible
to revert the translation process. Thus, the original content could be recovered by
performing a reversed translation (from the final ontology into the source one).
Those translations typically require previous ontology alignment.
Additionally, there are several recommendations and good practices for facilitating the achievement of semantic interoperability among systems [23] and reducing
the heterogeneity of information models for a same domain:
• reuse of existing ontologies at the possible extent: reuse of existing knowledge,
avoiding the construction of heterogeneous models that hinder interoperability. It
also implies to use recommended core ontologies in order to create a new ontology
covering the system needs, such as SSN in IoT domains.
• creation of ontologies following best practices [26].
• update and maintenance of ontologies [23].
• creation of catalogs of ontologies for easing the reuse of ontologies, and also their
update and maintenance or the creation of new ontologies aligned with appropriate
and core ones [23, 27].
The accomplishment of these recommendations makes more feasible that systems
follow a common ontology and share common semantics. Also, it significantly