measured on a continuous scale from 0-90. Our system must
allow its users to query the federation by their own vocabulary.
The system is then transparently handling and resolving the
discrepancies. The approach introduced in this chapter is called
semantic data sharing of spatial data.
Building semantics onto the syntactic description of geographic
objects can be considered as wrapping them with semantic
descriptors. Users are then interacting with the federation
through this semantic wrapper while the system is transparently
resolving the syntactic differences. However, in order to
establish such semantic descriptors, data sharing concept poses
some problems in several disciplines which have to be
resolved:
1. A common vocabulary must be defined that allows various
decision levels to exchange information at the semantic
level. This common vocabulary is know as
2. A set of protocols must be established that permit semantic-
level exchange of information. :
3. An architecture which implements the concept of semantic
data sharing.
4. Applying this concept on a large scale will definitely
increase the traffic on the network. A set of basic facilitation
services is required that off-load functionality such as name
service, buffering, routing of messages, and matching
procedures and consumers of information.
In this paper a mechanism for building a canonical data model
is introduced. Moreover a protocol for semantic-level exchange
is established. The system architecture and the network aspects
are outside the scope of this paper. An overview of the current
technology and research activities in the field of data sharing is
shown in section 2. The spatial canonical data model is
explained in section 3. The proposed concept for semantic data
sharing is shown in section 4.
2. LINKING HETEROGENEOUS SPATIAL
DATABASES
[Saltor et al., 1993] provided a comprehensive classification of
heterogeneity. The classification has three aspects: syntactic,
schematic, and semantic.
2.1 Syntactic Heterogeneity
Each database may be implemented in a different DBMS with a
different data model, e.g., relational model Vs object oriented
model. Syntactic heterogeneity is also related to the geometric
representation of geographic objects, e.g., raster and vector
representations.
Current technology and research activities for sharing spatial
information are tackling the above two aspects of syntactic
heterogeneity. One of the most prominent and promising
technologies which aim to provide connectivity between
heterogeneous databases is the open database connectivity,
ODBC [Kyle Geiger, 1995]. It can be plugged in most of the
current platforms. The main objective of ODBC is to resolve
the heterogeneity of the DBMSs, i.e., syntactic heterogeneity.
Users are able to interact with different platforms regardless of
their underlying operating system and DBMS. It is a standard
application programming interface (API) for accessing data in
both relational and non relational database management
systems. Using ODBC’s API, applications can access data
stored in a variety of personal computer, minicomputer, and
60
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B3. Vienna 1996
mainframe DBMSs, even when each DBMS uses a different
data storage format and programming interface.
The open GIS consortium, OGC is responsible for improving
the other aspect of syntactic heterogeneity, i.e., geometric
representation of geographic objects, using its open
Geoinformation specifications, OGIS [Schell D., 1995]. The
specifications have two parts: 1) the Open Geodata Model,
OGM, which provides a common geodata model for all spatio-
temporal data. The model supports both object and field based
approaches. 2) Open Geoprocessing Services, OGS. It defines
a common consistent set of geoprocessing software interfaces.
These interfaces define the behaviour of geoprocessing
software services which access, interchange, manage,
manipulate and present geospatial data specified in OGM.
The basic strategy of OGIS is to define a set of well known
types and common aggregates as the basic building blocks. The
well known types would include common programming types
such as integers, real numbers, character strings. The aggregate
types would include common programming database aggregate
constructors such as list, set, mullet-set, and tuple. Moreover
OGIS defines a basic level of spatial and temporal primitives
which would allows systems to build their own internal
representation. This includes point, line, areas, surfaces,
curves, and simplexes.
Provided that in the near future all GISs are using these
concepts in their basic definition, the transformation from one
spatial domain to the other is straight foreword process.
2.2 Schematic Heterogeneity
Objects in one database are considered as properties in the
other. Moreover, object classes of the same real world entity
may have different hierarchies and descriptors in different
databases. Unified data models are designed to handle this type
of heterogeneity. The concept proposed in this paper is
designed to handle the schematic and the semantic
heterogeneity simultaneously, section 4.
2.3 Semantic Heterogeneity
A real world entity may have been represented in different
ways by different designers in order to serve various
applications, giving as a consequence semantic conflicts at the
level of federation. For example a road network in a GIS for
transportation has different semantics from that in a GIS for
topographic mapping.
In the context of providing data sharing at the semantic level
[Daruwala A., et al., 1995] proposed a strategy based on the
notion of context interchange. In the context interchange
framework, assumptions underlying the interpretations
attributed to data are explicitly represented in the form of data
contexts with respect to a shared ontology [Goh C., et al.,
1994, 1995]. Ontology is a specification of a conceptualisation.
That is, an ontology is a description of the concepts and
relationships that can exist for a component GIS or a set of
interrelated components.
The ontology of certain application domain is implemented in a
component called mediator. A mediator is a paradigm which
provides a link between data sources and receivers [Siegel M.,
et al., 1991].
be realised b;
1. How dor
represente
2. How thes
detection
different :
A context r
receives a c
compares th
determine i
concessions
to as confl
conflicts are
undergoes ai
of forms: a
referred to i
sources exec
obtained foi
and converts
There are tw
1. All sour
explicitly
i.e., expe
2. All quer
mentione
It is worth
concept is
merging the
will result :
next sectio
Furthermor
modelling «
presented. '
defining th:
geographic
objects for
mediator se
for identify:
3. A
The propo:
building la
geographic
definition :
object bas
topologic t
and the ha
[Molenaar
introduces
generalisati
focused on
introduced
spatial obj
syntactic p
specificatic
A node is
divided in
context. A
way in wh
