International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B7. Istanbul 2004
emergency of situation, Status of the device, etc.) and
accordingly prepare the presentation. In this respect at
database level, a variety of algorithms have to be
developed: for generalisation and compression of graphics
and images (with respect of the screen resolution),
graphics-to-text (voice) conversion, etc.
e Semantic domain models (ontology based, see Section 4)
and translators between data from different sources and
domains that is reasonable to be implemented at database
level (see also next section).
e Preparation for distributed environments (open standards,
shared GII within INSPIRE and NGII), which are
composed of autonomous and heterogeneous components
(based on agreed interface specifications).
3.3 3D data in large scale outdoor/indoor models
One aspect, different in Disaster Management compared to
other applications of geo-information, is that the requirements
for combined indoor (internal plans of buildings and
construction) and outdoor (more traditional geo-information,
such as topographic, utility and cadastral data) models. In
general these models are large scale (1:500 or larger) and have
a true 3D nature. The systems responsible for creating and
managing these data sets are often quite different. For example
CAD system are used for creating indoor models of buildings
and constructions and GIS for outdoor geo-information.
However, users in urgent disaster management situations need
and expect seamless access to the information.
In order to realize this a number of problems have to be solved
(van Oosterom et al, 2004). The first one is bridging the
semantic gap between these different worlds (design versus
surveying). This is further discussed in Section 4.2. Once an
agreed model (covering aspects of the different world, CAD
and GIS) is created, different views on this representation may
be defined. The integrated GIS-CAD model is managed in a
way that consistency is maintained (during updates or adding
new data). The result will be that different applications may be
used to perform specialized tasks. This also implies that
different users may be working, at the same time, in different
environments (or at different locations) with the same model.
Similarly for both GIS and CAD worlds a gradual move from
file based approaches to DBMS approaches has occurred in
situations were the geo-information use has become more
structural and by more than a single individual (owning the
data).
Data management will benefit from well known advantages of
DBMSs: multiple user support, transaction support, security and
authorization, (spatial) data clustering and indexing, query
optimizing, distributed architectures, support the concept of
multiple views, maintaining integrity constraints (especially
referential integrity, but also other types). In summary, ‘island’
automation is abandoned and society wide information
management becomes a reality. The DBMS can be considered
an implementation platform for an integrated. model (with
different views). However, when exchanging information (or
using services from other sources), the structured exchange of
information becomes an important issue. The UML (also see
section 4.2) models are both the fundament for the storage data
models (further described in the DDLs of the DBMS) and the
exchange data models. The eXtensible Markup Language
(XML) can be used for the models at class level (XML schema
document ‘xsd’) containing the class descriptions and also for
the data at the object instance level (‘normal XML document
with data ‘xml’). XML documents also include the geometric
aspect of objects (examples are LandXML, GML, X3D, etc.)
4. DATA INTEGRATION AND KNOWLEDGE
DISCOVERY
The integration of multiple systems and databases is a common
necessity in large organisations. For disaster management it is
becoming a critical issue.
4.1 Problems in data integrations
There are three basic strategies for accessing data from multiple
sources: centralization, federation and collaboration (dynamic
integration). With centralization, a central data warehouse is
created to contain a copy of all or part of the information stored
in local database and managed by separate organizations (e.g.: a
central database contains the data from Police and Municipality,
copied from their respective databases). Frequent updates
ensure currency of the central database. Federation strategies
privilege access to multiple databases from a central location
without need for creating a central database. This strategy is
based on communication and connectivity, rather than content
centralisation. However, federation assumes an overall model
(or schema) of which the different distributed DBMS cover
parts of the content. In the third strategy, dynamic
collaboration, no such accepted overall (distributed) schema
exits. This strategy is based on dynamic data discovery and use
of the relevant sources. However, as the sources are
heterogeneous ‘translation services’ (for example in mediators)
are required for meaningful coupling and integration of the
information in applications. There are pros-and-cons in all three
solutions, but experience suggests that federation or dynamic
collaboration of content have the largest chances of success.
In a case of emergency, ‘mountains’ of data (static & dynamic)
are available and have to be analysed whether they are
useful/necessary for the current scenario or not. Analysing such
volumes of data clearly overwhelms the traditional manual
methods of data analysis such as spreadsheets, ad-hoc queries
and dedicated scripts. These methods can create informative
reports from data but cannot analyse the contents of those
reports to focus on important knowledge. There is an urgent
need for new methods and tools that can intelligently and
automatically transform data into information and, furthermore,
synthesize knowledge. These methods and tools are the subject
of the emerging field of knowledge discovery in databases.
There are at least three distinct cases in which information may
be lost when communicating between different language
groups, and by analogy, between Information Communities:
e [n the first case, definitions and concepts are shared but
there is no common language between the two groups, or
the groups share a common language but use dramatically
different dialects. This problem is corrected through
simple translation using a bi-directional mapping between
the two languages. As long as the languages themselves
are stable and there is a 1:1 relationship between relevant
terms this mapping solution supports effective
communication. For instance, if A and B want to talk
about build-up areas and the overland transportation of
rescue teams and A refers to ‘houses’ and B knows the
large constructions as ‘buildings’, they can agree that the
mapping of ‘house’ = ‘building’ will be used to
communicate this concept.
e In the second, somewhat more abstract case, a stable base
of definitions for terms is not shared between the
630
Interna
mit
Cc
re
efi
dif
4.2 Fo
The di
solving
include
pieces
purpose
such as
modelli
it becoi
(or eve
Also, s
importe
necessa
been m
ontolog
formali
At this
used re
of the
especia
(before
(familiz
differer
domain
informe
the sp
knowle
Structur
modelle
with c
part/wh
operatic
UML c
2002, ¢
become
framew
accurac
further
Object
OMG, :
valid m
classes
Besides
handlin
ontolog
