328
R. Gonzalez-Usach et al.
reduces the effort required for applying methods for the enablement of semantic
interoperability. It simplifies the process of ontology merging and ontology alignment
(due to the fact that ontologies share more commonalties). Moreover, the fulfilment
of these recommendations also helps to lower the complexity of semantic translations
and to improve their quality.
2.2 Semantic Interoperability in IoT
The achievement of IoT semantic interoperability faces significant challenges [23]
and remains one of the major challenges in IoT [2, 12].
First of all, IoT systems rarely follow common semantics. There is indeed a
tendency that each project or system creates their own ontology [23, 27, 28] and does
not seek consensus with other ontology developers [29], increasing the heterogeneity
of the representations of the information.
Second, in general the good practices mentioned in previous section are not
followed, adding complexity to potential enablers of semantic interoperability. Most
IoT systems do not follow existing ontologies, and new ontologies generally do
not reuse existing valid and well-established models. Regarding this point, W3C
Semantic Sensor Network (SSN) ontology is considered the standard for generic
modelling in IoT, and new ontologies should extend it adding the extra concepts
needed in a particular IoT area [23, 30]. The use of a well-consolidated ontology such
as SSN avoids to employ instead several ontologies for describing the same concepts
[31]. However, despite of these recommendations, there is a lack of methodologies
for modelling ontologies that fulfill the specific requirements for IoT data.
There is also an important need for ontology catalog creation barely covered,
although the IoT community has manifested some concern for building catalogs of
ontologies for some specific application domains [23, 27].
A third challenge is the heterogeneous formalism employed in IoT ontologies
and vocabularies. Most vocabularies are described in markup languages or defined
through UML artifacts; in most cases the formalism is not OWL. This fact hampers
data integration from common vocabularies [23].
Another significant challenge encountered is the current high difficulty on
unifying models, vocabularies and ontologies for semantically annotating the data.
There is an important necessity of tools and approaches able of providing a unified
semantic model aligned with vocabularies from IoT platforms [28].
In addition, due to the highly evolving nature of IoT, new IoT artifacts and features
are constantly appearing, and ontologies require fast and timely updates [23, 27, 28],
a fact that represents another challenge for IoT interoperability. By extension, also
ontology catalogs should be timely updated and maintained [23, 27].
Moreover, it has to be considered that IoT data management requires the ability
of handling large amounts of data (IoT big data) in real time, which requires
special processing capabilities on the devices that handle it for performing semantic
R. Gonzalez-Usach et al.
reduces the effort required for applying methods for the enablement of semantic
interoperability. It simplifies the process of ontology merging and ontology alignment
(due to the fact that ontologies share more commonalties). Moreover, the fulfilment
of these recommendations also helps to lower the complexity of semantic translations
and to improve their quality.
2.2 Semantic Interoperability in IoT
The achievement of IoT semantic interoperability faces significant challenges [23]
and remains one of the major challenges in IoT [2, 12].
First of all, IoT systems rarely follow common semantics. There is indeed a
tendency that each project or system creates their own ontology [23, 27, 28] and does
not seek consensus with other ontology developers [29], increasing the heterogeneity
of the representations of the information.
Second, in general the good practices mentioned in previous section are not
followed, adding complexity to potential enablers of semantic interoperability. Most
IoT systems do not follow existing ontologies, and new ontologies generally do
not reuse existing valid and well-established models. Regarding this point, W3C
Semantic Sensor Network (SSN) ontology is considered the standard for generic
modelling in IoT, and new ontologies should extend it adding the extra concepts
needed in a particular IoT area [23, 30]. The use of a well-consolidated ontology such
as SSN avoids to employ instead several ontologies for describing the same concepts
[31]. However, despite of these recommendations, there is a lack of methodologies
for modelling ontologies that fulfill the specific requirements for IoT data.
There is also an important need for ontology catalog creation barely covered,
although the IoT community has manifested some concern for building catalogs of
ontologies for some specific application domains [23, 27].
A third challenge is the heterogeneous formalism employed in IoT ontologies
and vocabularies. Most vocabularies are described in markup languages or defined
through UML artifacts; in most cases the formalism is not OWL. This fact hampers
data integration from common vocabularies [23].
Another significant challenge encountered is the current high difficulty on
unifying models, vocabularies and ontologies for semantically annotating the data.
There is an important necessity of tools and approaches able of providing a unified
semantic model aligned with vocabularies from IoT platforms [28].
In addition, due to the highly evolving nature of IoT, new IoT artifacts and features
are constantly appearing, and ontologies require fast and timely updates [23, 27, 28],
a fact that represents another challenge for IoT interoperability. By extension, also
ontology catalogs should be timely updated and maintained [23, 27].
Moreover, it has to be considered that IoT data management requires the ability
of handling large amounts of data (IoT big data) in real time, which requires
special processing capabilities on the devices that handle it for performing semantic