Semantic Web and IoT
15
the recorded activities, the type of data and the extracted features. Some classification
algorithms are found to perform very well in the majority of such studies, like Support
Vector Machines (SVM), Naive Bayes (NB) and Decision Trees [94]. If a system
consists of different sensors and there is a need to utilize the information provided by
all of them, fusion methods are applied. Fusion is the combination of information and
that can be performed after the classification process to combine the classification
results of each sensor or at earlier stages, before the data enter a classifier, where it
combines the extracted features of different sensors [95].
3.2 Modelling Multi-modal Events and Observations
3.2.1 Location
Data semantics are extensively used in location-based services (LBS), in order to
find and integrate the information related with the users. Several LBS were analyzed and recorded in [96]. First they have classified the LBS data based on relevant
definitions and use. A distinction was made between Domain Data, Content data
and Application data where the Domain Data include spatial and temporal concepts
(e.g. location, position, time etc.), Content data mainly describe specific content, and
finally, Application data comprising of the actual services and user profiles.
3.2.2 Activities—Events
Event-Model-F [97] describes a process for identifying and describing real word
events. It is based on DUL and follows the ”descriptions and situations” (DnS)
ontology design framework [98] for modelling different concepts of events, such
as object attendance, relationships, and different meanings of the same event by
introducing six ontology design patterns. In addition to the DnS model, Event-ModelF implements a number of internal representations to describe relationships among
events, such as causality and correlation. Figure 1 [99] describes the pattern of EventModel-F correlation of Events.
The Simple Event Model Ontology (SEM) [100] is an attempt to establish an
ontology model for events with no extreme semantic restrictions. The open nature
of the Web itself and the necessity to design various perspectives of the same event,
support this decision. The proposed ontology has core classes such as Event, Actor,
Place and Time and corresponding properties that allow us to model fundamental
facts. This also involves means to express some restrictions related to different points
of view, namely: (1) Event bounded roles, (2) time bounded validity of facts (e.g. type
dependent type or roles) and (3) attribution of the authoritative source of a statement.
15
the recorded activities, the type of data and the extracted features. Some classification
algorithms are found to perform very well in the majority of such studies, like Support
Vector Machines (SVM), Naive Bayes (NB) and Decision Trees [94]. If a system
consists of different sensors and there is a need to utilize the information provided by
all of them, fusion methods are applied. Fusion is the combination of information and
that can be performed after the classification process to combine the classification
results of each sensor or at earlier stages, before the data enter a classifier, where it
combines the extracted features of different sensors [95].
3.2 Modelling Multi-modal Events and Observations
3.2.1 Location
Data semantics are extensively used in location-based services (LBS), in order to
find and integrate the information related with the users. Several LBS were analyzed and recorded in [96]. First they have classified the LBS data based on relevant
definitions and use. A distinction was made between Domain Data, Content data
and Application data where the Domain Data include spatial and temporal concepts
(e.g. location, position, time etc.), Content data mainly describe specific content, and
finally, Application data comprising of the actual services and user profiles.
3.2.2 Activities—Events
Event-Model-F [97] describes a process for identifying and describing real word
events. It is based on DUL and follows the ”descriptions and situations” (DnS)
ontology design framework [98] for modelling different concepts of events, such
as object attendance, relationships, and different meanings of the same event by
introducing six ontology design patterns. In addition to the DnS model, Event-ModelF implements a number of internal representations to describe relationships among
events, such as causality and correlation. Figure 1 [99] describes the pattern of EventModel-F correlation of Events.
The Simple Event Model Ontology (SEM) [100] is an attempt to establish an
ontology model for events with no extreme semantic restrictions. The open nature
of the Web itself and the necessity to design various perspectives of the same event,
support this decision. The proposed ontology has core classes such as Event, Actor,
Place and Time and corresponding properties that allow us to model fundamental
facts. This also involves means to express some restrictions related to different points
of view, namely: (1) Event bounded roles, (2) time bounded validity of facts (e.g. type
dependent type or roles) and (3) attribution of the authoritative source of a statement.