- (D 
ON THE ACQUISITION, REPRESENTATION AND APPLICATION OF KNOWLEDGE 
IN GEO-INFORMATION SYSTEMS 
Ralf Bill, Senior Researcher, Institute of Photogrammetry, University of Stuttgart, Federal 
Republic of Germany, Commission III 
30th April 1992 
ABSTRACT 
The paper presents the concepts of knowledge acquisition, rep- 
resentation and application with respect to their usage in geo- 
information systems (GIS). Recent database research activities 
are partially integrated into commercially available GIS-products 
whereas the use of knowledge administration and processing on 
the other hand is far away from any development in the GIS- 
community. Examples should illustrate their general applicabil- 
ity in the GIS-area. 
Beginning with some definitions on data, information and knowl- 
edge various forms of knowledge are mentioned. The knowledge 
base is the major part of a knowledge-based GIS; there the rules 
of type 'If - then' are stored. In GIS-products one can find knowl- 
edge about rule-based relations and prototypical knowledge inte- 
grated in a procedural manner. The integration of fuzzy knowl- 
edge would be usefull for certain applications. 
The knowledge representation formats are treated in detail, they 
are logic models, production rules, semantical networks and 
frames. Similarities to database research activities such as the 
entity-relationship-model and object-oriented are discussed. 
The architecture of a GIS differs very much from the demand 
of a knowledge-based system. The procedural design of a GIS 
needs to be replaced with the inference-mechanism of an expert 
system, heuristical approaches instead of procedural ones are to 
be integrated. 
Different GIS-applications of knowledge integration such as fuzzy- 
polygon overlay, rule-based raster-vector conversion, object mod- 
elling based on semantic networks or frames are presented and 
the advantages illustrated. 
KEY WORDS : Knowledge acquisition, knowledge represen- 
tation, knowledge-based systems, GIS 
1 INTRODUCTION 
1.1 Definitions 
At the beginning of this paper the major terms 'data, information 
and knowledge’ should be defined. It is not the intention of the 
author to define these in an overall context, because this would 
end in a philosophical discussion. So, the given definitions are 
context-related. Figure 1 tries to illustrate the relation between 
data, information and knowledge in a simplified manner. 
Data in the classical meaning are simply characters. These data 
can be read, stored, compared, processed and written by a com- 
puter. Extending the definition of character data could also be 
images, texts, language, too. Data itself are of meaning for a 
735 
computer but without any knowledge about the meaning and 
structure of the data a human being is not able to read and in- 
terprete it easily. 
Information is related to data, but we want to see it here as a 
result of using transformations, rules and knowledge familiar to 
those working with the data to gain facts and interpretable re- 
sults in a given context. Knowing the way data are organized in 
a computer and the purpose of the data storage in the computer 
a human being is able to work with the data. Information is to 
be differentiated in three levels; the syntax, i.e. the characters 
used, the semantics, i.e. the meaning, and the communication, 
i.e. the way information is distributed. 
The knowledge of a knowledge carrier is defined as the sum of all 
declarations about the represented part of the world regarded as 
true, that are really true. Persuasion in comparison to knowledge 
is all what the knowledge carrier believes to be true. 
1.2 Spatial information systems 
Spatial information systems (or geographical or geo-information 
systems (GIS) as synonyms) are computer-based systems which 
consists of hardware, software, data, and the applications. A spa- 
tial information systems supports the digital input, management, 
analysis, and presentation of spatial data (R. Bill and D. Fritsch 
(1991), P. A. Burrough (1986)). Spatial or Geo-Information sys- 
tems are combining the database technology with the information 
system technology under the common aspect of treating spatial 
phenomena. For both parts of technology this is not to be seen as 
a standard problem. In computer science only a small group of 
researchers is currently working on spatial problems. This led to 
the situation we have today, that a large gap exists between the 
research domain in computer science and the development and 
application area of GIS. Only a small number of GIS-products 
currently available on the market is close to the research activ- 
ities e.g. in object-oriented databases. The situation becomes 
even worse if we look at the topic of this paper, the integration 
of knowledge in a GIS. 
Data in a GIS Currently available GIS-products are dealing 
with the following data types : 
e geometry : the coordinates and their metric. Geometric 
primitives are points, lines and surfaces in 2 or maximum 
2.5 dimensional way. 
e topology : the relation of geometric primitives and objects 
in itself and to others. Topologic primitives are nodes, edges 
and meshs. Their relations are treated in sharing, connec- 
tivity, neighbourhood, inclusion etc. 
attributes : information describing the non-geometric prop- 
erties of the objects.