Information Technology Reference
In-Depth Information
Semantic-Based Context Modeling
The common approach is that SOUPA (Chen et
al. 2005) is used as a starting point extended for
the needs of the application field. Our first step
towards a holistic view of context awareness in
smart spaces has been introduced in (Toninelli et
al. 2009; Pantsar-Syväniemi et al. 2010) by map-
ping the dimensions of context to the levels of the
context defined in (Bettini et al. 2009). To avoid
“yet another” definition of context based on the
kind of information that it conveys, our approach
is to shift the focus from the content to the purpose
of the context. Instead of trying to describe all of
the possible types of context data that might be
of interest to the smart applications, we assume
that any piece of data might be the context for
a given application (and possibly not for other
applications). Context is strongly application-
specific: the same piece of information cannot be
defined a priori as “the context” unless this notion
refers to a specific smart application at a specific
time. Thus, defining the context in smart spaces
is more about how, why, and by whom the smart
space-related information is used, rather than about
what the information describes. Given that, we
propose that “A context defines the limit of the
information usage of a smart space application”.
The notion of “information usage” is intended to
be as comprehensive as possible, and includes
the retrieval, access, understanding, processing,
production, and sharing of information by smart
applications.
As a consequence of this approach, the context
interoperability finally boils down to data interop-
erability, since the context itself is represented
as information in the smart space. Therefore,
the same semantic representation of data that
ensures interoperability at the information level
also supports the meaningful exchange of context
across smart applications. This approach helps to
understand how the context data is to be dealt with
within the physical context, in order to achieve
pragmatic interoperability (see Table 1). After
that, the context data is enhanced with additional
context data at the second level that is responsible
Compared to other approaches, ontological models
for context and context-aware adaptation strate-
gies provide a clear advantage, both in the terms
of heterogeneity and the interoperability of data,
context and context changes. However, there is
very little support for modeling temporal aspects
in ontologies, and reasoning with ontologies based
on Description Logics (e.g., OWL-DL) may pose
serious performance issues. Semantic technolo-
gies have been applied to support the context-
awareness in several emerging smart spaces and
pervasive computing platforms, including (Chen
et al. 2004, Wang et al. 2004) and many others.
Most systems exploit semantic techniques to
represent and reason context and adapt service/
application behavior accordingly. Ontologies
have also been developed for describing QoS,
but a lack of completeness is common to all of
the approaches; only one or a few qualities are
considered, and the vocabulary or/and metrics
are missing. Moreover, making tradeoffs between
quality attributes and managing QoS at run-time
are not supported. Concerning the contextual
characteristics of services, several ontologies
have been designed, some of which are more
elaborate and others more succinct, depending on
their scope. Most of the approaches address the
vocabularies of pervasive computing. Typically,
they include a set of vocabularies for describing
people, agents, and places, as well as a set of
properties and relationships that are associated
with these basic concepts. However, rather little
emphasis has been placed on temporal contextual
information. Moreover, no attempts have been
made to align the service and context ontologies.
Given the best variety of existing context
ontologies, the following challenge is how to pro-
vide a suitable definition of the context for smart
spaces. A major obstacle is that the set of context
ontologies that have been proposed for pervasive
computing environments has not been standard-
ized nor widely accepted and systematically used.
Search WWH ::
Custom Search