Information Technology Reference
In-Depth Information
location and between walking aid and location.
Entities handled in L2 can also be seen as
situa-
tion objects
.
L3 (
Impact Assessment
) refers to the predic-
tion of the occurrence of certain situations and
its probability. Related to the example this means
to find out if the prerequisites for the described
situation can come true. It is clear that the person
has to be in the first floor to fall down the stairs.
If the person is actually somewhere in the base-
ment, it is not likely that the defined situation
occurs. Whereas if the person is in the first floor
and moving towards the stairs, the probability
of the critical situation to come true increases.
There are more sophisticated descriptions which
allow a precise definition for this use case, but it
is a sufficient example to illustrate the JDL Data
Fusion Model and its potential usage for building
a situation aware AAL system. In the next section
mechanisms for modeling and describing such
situations will be discussed in general.
events and defined context. PERSONA (Amoretti,
Wientapper, Furfari, Lenzi, & Chessa, 2009) is
another project from the AAL domain and mainly
focuses on activity monitoring. It uses a
context
ontology
to process incoming camera data. The ap-
proach of using ontologies for modeling situations
is also motivated by various researchers beyond
the AAL domain. (Baumgartner & Retschitzeg-
ger, 2006) use ontologies to define dangerous
situations in the field of road traffic management.
(Matheus, Kokar, & Baclawski, 2003) describe
various ontology-based approaches, compare their
advantages and disadvantages and use them to
describe battlefield scenarios and supply logistics.
(Kokar, Matheus, & Baclawski, 2009) have shown
that ontologies can be used to create computer
processable semantics that are compatible with
Endsley's model of human situation awareness
as well as with the situation theory of Barwise.
AAL systems have to fulfill multiple require-
ments. The detection and treatment of emergency
situations is the most important one of the require-
ments. To provide reliable emergency detection
two major questions need to be solved:
(ii)
how to
describe a situation of interest (i.e. a critical situ-
ation), and
(ii)
how to find out when this situation
has happened, or, if possible, is about to happen?
The first question is answered by the usage of
an ontology that provides the basic notations for
defining objects, attributes and relations between
the objects. This vocabulary is used to describe a
situation for a specific domain (Baumgartner &
Retschitzegger, 2006) and can be defined outside
the AAL domain in an own ontology. Therefore,
it is possible to encapsulate different concepts
and facilitate knowledge sharing among related
domains (Happel & Seedorf, 2006). To define a
critical situation, domain experts need to describe
relations among objects. An example of such a
reasoning expression can be found in the previ-
ous section Situation Awareness and Situation
Detection. To answer the second question - find
out when this situation has come true - it has to
Situation Modeling and Description
To define and store data in a machine process-
able form, a model is needed. The development
of such a model is a challenging task and has to
cope with various issues. These involve differ-
ent data sources and objects, data acquisition
and storage strategies as well as data representa-
tion and the definition of relations between data
objects. According to the survey of (Strang &
Linnhoff-Popien, 2004), various approaches like
(ii)
markup scheme models, based on hierarchical
data structures,
(ii)
graphical models, based on
UML or ORM,
(iii)
object oriented models,
(iv)
logic based models using expressions and rules
or
(v)
ontology based models, exist.
Ontologies are a formal representation of
knowledge and allow to describe concepts and
relations between the concepts. The SOPRANO
(Wolf, Schmidt, & Klein, 2008) project, for ex-
ample, uses ontologies to describe relations among
Search WWH ::
Custom Search