204
G. E. Modoni and M. Sacco
not). In the recent years, the efforts of researchers are also addressed towards stream
reasoning, which aims to process dynamic (data in motion) data, unlike traditional
reasoning which processes static data [21].
3.3.7 Databases Federation and Replication to Support Data
Distribution
Distributed approaches based on paradigms such as fog and edge Computing offer
valid mechanisms to distribute data and intelligence among the various software and
hardware components of an internet of things based architecture (microcontrollers,
sensors, services, and databases on the cloud, etc.), thus offering more suitable methods for respond to the uninterrupted rise of the number of devices which more and
more are connected to the network. Leveraging these distributed approaches, it is
possible to transfer part of the computing power and storage close to the sources of
data, thus improving the bandwidth consumption and increasing the data availability
In order to identify a scalable architecture model based on these distributed
paradigms, one significant contribution can come from the implementation of a
federation of different RDF stores (also called virtual integration) which allow to
perform complex and structured queries against a federated set of data sources [14].
In fact, the federated approach goes into the direction of a distributed approach such
the one that offered by fog and edge computing. It should be underlined that the
federation query processing is enabled by the nature of linked data inherent in the
semantic models managed by the RDF stores. The federation of the databases can
be also combined with their (partial) synchronization replication, eventually limited
to some specific data models [1], thus organizing a peer-to-peer network where different autonomous actors can collaboratively update and enrich the knowledge base
stored in the semantic datasets.
3.3.8 Handling Spatio-temporal Data
A growing number of software applications generate spatiotemporal data which
track the position of moving objects (e.g. cars, etc.) or moving people. To integrate
spatiotemporal data with other relevant information coming from other sources, it can
be useful to represent them under the form of RDF and stored in the RDF stores [34].
The spatiotemporal data are characterized by the following two significant features:
(a) the values of these data can continuously change, also with a high velocity; (b) the
volume of these data is high. Despite spatiotemporal data processing is a consolidated
feature, it is not easy to integrate it with RDF data management and for this reason
efforts of researchers must mainly addressed towards scalability issues to manage
large scale and dynamic spatiotemporal RDF data [34].
G. E. Modoni and M. Sacco
not). In the recent years, the efforts of researchers are also addressed towards stream
reasoning, which aims to process dynamic (data in motion) data, unlike traditional
reasoning which processes static data [21].
3.3.7 Databases Federation and Replication to Support Data
Distribution
Distributed approaches based on paradigms such as fog and edge Computing offer
valid mechanisms to distribute data and intelligence among the various software and
hardware components of an internet of things based architecture (microcontrollers,
sensors, services, and databases on the cloud, etc.), thus offering more suitable methods for respond to the uninterrupted rise of the number of devices which more and
more are connected to the network. Leveraging these distributed approaches, it is
possible to transfer part of the computing power and storage close to the sources of
data, thus improving the bandwidth consumption and increasing the data availability
In order to identify a scalable architecture model based on these distributed
paradigms, one significant contribution can come from the implementation of a
federation of different RDF stores (also called virtual integration) which allow to
perform complex and structured queries against a federated set of data sources [14].
In fact, the federated approach goes into the direction of a distributed approach such
the one that offered by fog and edge computing. It should be underlined that the
federation query processing is enabled by the nature of linked data inherent in the
semantic models managed by the RDF stores. The federation of the databases can
be also combined with their (partial) synchronization replication, eventually limited
to some specific data models [1], thus organizing a peer-to-peer network where different autonomous actors can collaboratively update and enrich the knowledge base
stored in the semantic datasets.
3.3.8 Handling Spatio-temporal Data
A growing number of software applications generate spatiotemporal data which
track the position of moving objects (e.g. cars, etc.) or moving people. To integrate
spatiotemporal data with other relevant information coming from other sources, it can
be useful to represent them under the form of RDF and stored in the RDF stores [34].
The spatiotemporal data are characterized by the following two significant features:
(a) the values of these data can continuously change, also with a high velocity; (b) the
volume of these data is high. Despite spatiotemporal data processing is a consolidated
feature, it is not easy to integrate it with RDF data management and for this reason
efforts of researchers must mainly addressed towards scalability issues to manage
large scale and dynamic spatiotemporal RDF data [34].