ul 2004
ometric
etc.).
;E
'ommon
ent it is
multiple
]ynamic
1ouse is
n stored
s(e.g.:a
cipality,
updates
trategies
location
rategy is
| content
ll model
[S cover
dynamic
schema
/ and use
ces. are
ediators)
n of the
all three
dynamic
cess.
dynamic)
they are
sing such
| manual
c queries
formative
of those
an urgent
ntly and
thermore,
1e subject
latabases.
ation may
language
ties:
hared but
zroups, or
imatically
| through
x between
1emselves
n relevant
effective
nt to talk
rtation of
cnows the
e that the
used to
table base
ween the
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B7. Istanbul 2004
communities. Correcting for this requires a direct mapping
of shared definitions plus a set of interpretations for terms
that can't be mapped. Where there is a 1:M mapping of
definitions between communities, generalization and a
consequent loss of information will occur when mapping
multiple, specific definitions to one more general
definition.
Finally, there is the case in which basic concepts are not
shared between the communities, i.e., when the two
communities have starkly different worldviews. For
instance, consider the real world features ‘snow’ and
‘transportation systems’. Suppose one Information
Community recognizes only the first, and the second
recognizes only the second. Attempting to address the
effects of snow on transportation systems would be
difficult or impossible in either community.
4.2 Formal semantics
The discussion above clearly shows that an important key
solving the addressed problems is capturing the semantics
included in the different models. Implicit knowledge or nice
pieces of natural text and tables are not sufficient for this
purpose. A more formal approach, as developed in disciplines
such as knowledge engineering, ontology and object-oriented
modelling, is required. Based on this formal semantic approach
it becomes possible to decide whether different domain models
(or even models within one domain) are or can be harmonized.
Also, spatial information handling by machines will become
important, which makes the formal approach even more
necessary. In the last decade important technology progress has
been made in the discipline of knowledge engineering (UML,
ontology, semantic web), which enables further knowledge
formalization in a practical manner.
At this moment most spatial (both CAD and GIS) information is
used relatively direct by humans, in the future also large parts
of the information will (first) be processed by machine,
especially in time-critical situations of disaster management
(before communication again with humans). While a human
(familiar within a specific domain) is capable of interpreting
different concepts by using implicit context information (which
domain is under concern, who did supply/produce the
information, etc.), for a machine (or humans not familiar with
the specific domain) it will be necessary to make this
knowledge explicitly available. A large part of the formal
structural knowledge concerning the concepts (objects being
modelled) is captured in the relationships that an object has
with other types of objects (specialization/generalization,
part/whole, association), characteristics (attributes) and
operations (methods, functions) belonging to the object class.
UML class diagrams are often used for this modelling (OMG,
2002, chapter 3, part 5). The use of UML class diagrams has
become the ‘default’ approach when creating formal knowledge
frameworks, but the graphic diagram has a limited semantic
accuracy. Within UML a non-graphic language is provided for
further modelling semantics (knowledge frameworks), i.e. the
Object Constraint Language (OCL, see OMG, 2002, chapter 6,
OMG, 2003). This can be used to specify conditions to which a
valid model should adhere (constraints); such as invariants for
classes and pre- and post-conditions for operations.
Besides UML (and OCL) there are also specific tools for
handling (‘reasoning’) with formal concepts (semantics,
ontology); e.g. translation the terms/concepts from one domain
631
to the terms/concepts of another domain. Possible tools are
OWL, the Web Ontology Language (W3C, 2004) or the new
ODM (Ontology Definition Metamodel) development from the
OMG of which the final adoption is expected November 2004.
4.3 Further research
Some of the most important issues to be considered in
semantic/data discovery domain are:
e Integration of thematic, contingency and real-time data in
preparation for knowledge discovery and emergency
knowledge transaction processing.
e Developing context-aware engines and agents for query
and analysis with respect to the type of the front-end and
communication channels used.
* Investigation, adaptation and development of converters to
well-known Web standards and formats.
e Developing knowledge-based systems for browsing and
analysis in a distributed data environment.
e Investigating and developing intelligent semantic-based
engines and corresponding translators for semantic search
and analysis. :
S. POSITIONING OF MOBILE WORKERS AND USERS
As mentioned in Section 2, highest requirements are coming
from the mobile users and workers. To be able to discover the
most appropriate information, the system may need 3D
positions of the users. Furthermore the system has to be able to
maintain continuous communication related to both rescue
forces (Police, Ambulance, Fire Brigade) and citizens.
The required accuracy of the positioning may depend on the
case and may vary from 100 meters (locating a hospital) up to 5
meters (locating and safe exit in a building with reduced
visibility). The system should be able to detect what kind of
situation appears and selectively decide on the preferred way of
positioning and communication (depending on the availability
of networks).
5.1 Positioning
Several possibilities for positioning can be considered: Global
Navigation Satellite Systems (GNSS), telecom networks,
WLAN or hybrids of them. All approaches have advantages and
disadvantages (Zlatanova and Verbree 2003). :
At the moment the only available relatively low-cost Global
Navigation Satellite System devices offering 3D positioning
and navigation capabilities are GPS devices. Although these
devices are designed to track up to 12 satellites simultaneously,
they receive in dense build-up areas not that easy the minimal 4
satellite single frequency signals necessary to determine a 3D-
position. The configuration of these line-of-sights is not optimal
either, limiting the accuracy to less than the 10 meters, which
could be obtained with a clear view. At first sight, both
accuracy and availability are not suitable for the rapid and
precise positioning necessary for tracing ad tracking mobile
workers and users within disaster management application.
Furthermore, with a cold start, the receivers needs a certain
start-up time to acquire the satellite almanac, necessary to know
where to look for a certain satellite. Within buildings and other
closed spaces the satellite signal is too weak to use. Finally, the
receivers should be carried in such a way as the antenna is more
or less positioned to the sky. All these limitations are no in
favour of for using GPS as it is currently available.
