| on its
text car
| and at
ngle or
orapher
1 of the
tabases
entified
] in this
optimal
ile the
rmed in
specific
cess by
IN
making,
ological
process
ıportant
ologies.
ography
cussing
mobile
n maps
Distiller
r small
alent of
tion and
le scale
he most
j. ]t is
GIS and
tial data
irch and
ation of
several
rts and
consider
rapher.
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B4. Istanbul 2004
Spatial data obtained from the real world is generalized in two
steps: Model and Cartographic generalization. These are the
two main components of generalization process. Model
generalization is the simplification of the abstract digital model
represented by the geographic information and this stage
consists of no artistic and intuitive components (Kilpelainen,
1997). It is applied in database and considered as a
preprocessing stage for cartographic generalization. On the
other hand cartographic generalization consists of both of these
components as a complementary part of generalization process
so it is one of the reasons why cartography is considered as an
art. As a result, cartographic generalization has the leading role
in the transmission of the data by using symbols to represent
geographic reality and it is a significant stage of the map
production process.
2.1 Generalization Operators
Studies done by Shea and McMaster (1989) resulted with a
conceptual model for generalization process. They modeled this
process based on three main questions; why, when and how we
should generalize? First two questions include steps needed
while deciding generalization but third one is about utilization
step. Generalization operators are the answer of the third
question. Different authors name these operators as steps, tools
or processes of generalization. They are all correct because
these operators are the methods used to generalize data. Shea
and McMaster (1989) made a detailed definition of 12 operators
while answering the third question of their model. 10 of these
operators are defined for spatial transformations. They entitled
as simplification, refinement, smoothing, displacement,
amalgamation, exaggeration, aggregation, enhancement,
merging, and collapse. On the other hand rest of these twelve
operators, classification and symbolization, consist of attribute
transformations. Although the authors studying on special
issues ‘define different additional operators as Kilpelainen
(1997) did for MRDB system, these 12 operators form the basic
infrastructure of generalization process.
2.2 Needs for Generalization
Generalization has always played an important role in map
production. However the scale is an important and determining
concept for map contents so it is generally agreed that scale is
the most important constraint of the generalization (Bildirici,
2000). Another limitation for the generalization is the aim of
the map. In addition to the scale and aim of the maps, quality
and quantity of data and graphic limitations are considered as
the factors that affect generalization process by Robinson et al.
(1978). Moreover, Kilpelainen (1997) emphasized the effects of
the human factor, the cartographer, over the generalization
process by her research succeeded with Finnish cartographers.
As it is known, generalization process is a set of rules.
Especially these rule bases are very important for automation of
the process. Maintenance of the topological consistency during
the generalization is one of these rules. Topology is the
mathematical concept of spatial structure, sometimes defined as
“characteristics of geometry that do not change when the
coordinate space is deformed" (Hardy et al., 2003). In other
words, topology is a structure that defines geometrical
relationships between objects. Hardy et al. (2003) state that:
” Shared edges between land polygons,
- Junctions between streets in the road network,
- Colinearity of administrative boundaries with
roads and streams,
245
= Adjacency of buildings to roads,
need to be defined explicitly for good generalization. Moreover.
topology is very important for road networks. If any model is
tried to be set up for roads, first its topological relations should
be defined then this topological structure should be formalized
by using an appropriate method. However, Such an approach is
followed in this work.
3. MULTIPLE REPRESANTATIONAL DATABASES
(MRDB)
Although there is just one world reality, its representations vary
with different aims, contents or display scale so different level:
of representations of real world become a requirement for the
experts. This requirement is increased by the development o:
the technologies on GIS, which is an inter-disciplinary work
However, generally different representations are aimed as an
output in different GIS applications. Researches done foi
covering these kinds of needs resulted with the MRDB. The
National Center for Geographic Information and Analysis
began discussion of objectives and process of developing a
research agenda in MRDB in the late 1980s (Buttenfield
&Delotto 1989).
Multiple Representations are the different representations of the
same spatial database. These representations can be in different
scale, aim and resolution. MRDB is a spatial database, which
can be used to store the same real world phenomena at different
levels of precision, accuracy and resolution (Kilpelainen, 1997).
A comprehensive description of the MRDB is done in
Kilpelainen (1997) and she formed an MRDB model for
generalization of geo-databases for topographic maps.
According to the Kilpelainen’s model MRDB consists of three
main components: representation levels, connectivities and
reasoning process. Representation levels cover the base level,
which has the most detailed representation of the objects, and
higher levels in which object representations vary with the
scale, aim or resolution. Number of the higher levels change in
terms of application so they can be defined as application
dependent levels. MRDB aims to provide the propagation of the
updates applied on base level to the other representational
levels automatically so connectivities should be described and
formalized between objects in the same or different levels.
Kilpelainen (1995), separate the relations between different
objects at one level, relationships, from the relations between
the different representations of the same object at different
levels, connectivities. Finally, the reasoning processes are
needed to provide full functionality in the MRDB. It means that
the updates can be propagated from lower level representations
by using the model generalization operators applied
automatically in the modules to be generalized (Kilpelainen,
1997).
Today MRDB is one of the most important subjects of
concerning disciplines. Because this is a different database
approach developed to cover the current problems on data
management, automatic generalization and map production.
Because data sets and map series are obtained in different
European countries, many projects for adopting MRDB are
implemented and MRDB applications start to be expansive
during this adaptation process.
4. CASE STUDY
As Timpf et al. (1992) stated, navigation is a fundamental
human activity and an integral part of everyday life. People
