338
R. Gonzalez-Usach et al.
the common format to authorized users, while maintaining the security and privacy
of the system.
The Data Lake provides a common interface to retrieve the data as if all the
data were contained in the same database, to feed big data analytics methods to
enable predictive analysis, or feed applications built on top. Thanks to semantic
interoperability, the applications built on top of the Data Lake will receive the data
in the common ontology and data model, regardless of their origin.
Data Analytics
The Data Analytics back end contains components for the analysis of the data
retrieved from the Data Lake in order to extract meaningful information. These
methods include feature extraction, feature selection, anomaly detection, prediction,
clustering and hypothesis testing. The Data Analytics methods allow the analysis of
several types of parameters and the identification of patterns and expected values,
the comparison between users or the comparison of values from the same user at
different contexts and the detection of deviations.
Since the Data Analytics component obtains the data in the common ontology and
data model of ACTIVAGE, it offers services that are independent of the IoT platforms
that collected the data. The Data Analytics methods are exposed through the AIoTES
API and can be used by applications built on top of AIoTES. Moreover, these results
can be displayed in a comprehensible way using the visualization tools included in
the Service Layer of AIoTES, which also provide a graphical user interface for the
data analytics methods.
3.5.2 Platform-Independent Applications
The existing AHA applications have been built on top of a specific IoT platform.
This aspect limits the interoperability because the migration to a new platform would
require to adapting the applications to the syntax and semantics of the new platform. Thus, without a common syntax and semantics, the effort required to integrate
an application with a new DS or platform would increase exponentially. Similarly,
without AIoTES, the creation of new multi-platform or multi-DS applications would
be significantly more complex and prone to error, since they would need to communicate with different interfaces and deal with several syntactic and semantic representations of the data. For these reasons, the development of services that could be
adopted by any DS would not be feasible without AIoTES.
Thanks to the interoperability framework used in ACTIVAGE, the inclusion of
new platforms or DSs does not require any changes in the existing services and
applications. Moreover, the interoperability approach followed in ACTIVAGE also
enables the creation of platform-independent applications. As long as a common
format is used, applications can be built on top of AIoTES and be independent from
the platform and DS. These new applications can be uploaded to the ACTIVAGE
R. Gonzalez-Usach et al.
the common format to authorized users, while maintaining the security and privacy
of the system.
The Data Lake provides a common interface to retrieve the data as if all the
data were contained in the same database, to feed big data analytics methods to
enable predictive analysis, or feed applications built on top. Thanks to semantic
interoperability, the applications built on top of the Data Lake will receive the data
in the common ontology and data model, regardless of their origin.
Data Analytics
The Data Analytics back end contains components for the analysis of the data
retrieved from the Data Lake in order to extract meaningful information. These
methods include feature extraction, feature selection, anomaly detection, prediction,
clustering and hypothesis testing. The Data Analytics methods allow the analysis of
several types of parameters and the identification of patterns and expected values,
the comparison between users or the comparison of values from the same user at
different contexts and the detection of deviations.
Since the Data Analytics component obtains the data in the common ontology and
data model of ACTIVAGE, it offers services that are independent of the IoT platforms
that collected the data. The Data Analytics methods are exposed through the AIoTES
API and can be used by applications built on top of AIoTES. Moreover, these results
can be displayed in a comprehensible way using the visualization tools included in
the Service Layer of AIoTES, which also provide a graphical user interface for the
data analytics methods.
3.5.2 Platform-Independent Applications
The existing AHA applications have been built on top of a specific IoT platform.
This aspect limits the interoperability because the migration to a new platform would
require to adapting the applications to the syntax and semantics of the new platform. Thus, without a common syntax and semantics, the effort required to integrate
an application with a new DS or platform would increase exponentially. Similarly,
without AIoTES, the creation of new multi-platform or multi-DS applications would
be significantly more complex and prone to error, since they would need to communicate with different interfaces and deal with several syntactic and semantic representations of the data. For these reasons, the development of services that could be
adopted by any DS would not be feasible without AIoTES.
Thanks to the interoperability framework used in ACTIVAGE, the inclusion of
new platforms or DSs does not require any changes in the existing services and
applications. Moreover, the interoperability approach followed in ACTIVAGE also
enables the creation of platform-independent applications. As long as a common
format is used, applications can be built on top of AIoTES and be independent from
the platform and DS. These new applications can be uploaded to the ACTIVAGE