intelligence does not exist. At most, some
features which characterise intelligent beings
can be listed and on this basis the phenomenon of
intelligence can be understood. Therefore
human beings are only able to formulate a
definition of quasi-intelligence which remains
an approximation of the ideal of intelligence,
and in consequence to build quasi-intelligent
systems. This led to the title of the paper. Based
on the above arguments the term intelligence in
this paper will be identified with the term quasi
intelligence. One of the features which
characterises a truly intelligent system is its
ability to collect information about its
environment. An intelligent system can also
synthesise the collected data, draw conclusions
and make decisions. These powers of an
intelligent system would make it possible to
transfer some duties of the user to the system. It
would result in more effective and user-friendly
operation of the system.
The general feature of existing software
packages is their passiveness in the collection of
information about their environment. A
computer program is only able to use
information which an operator correctly inputs
to it. At best, if the input information is wrongly
formulated, the program will provide an error
message. At worst, the program will not detected
errors in the input and incorrect information
will be derived, or even incorrect decisions
made. All data introduced to the system, good or
bad includes some information. This
information can be used, for example, in order to
improve the communication conditions between
operator and the system. This is a very
important aspect, because the human operator is
the bottleneck in the human-system
configuration. Thus, moving some functions
from the human to the system should improve the
effectiveness of the work of the human-system
configuration.
The above remarks suggest an idea for
equipping the system with the capability to
observe the operator, for drawing conclusions on
the collected data and for decision making. A
system with these features can be described as
"active” with regard to its contact with its
environment.
Let us consider how we can improve the
effectiveness of the interaction between an
operator and a system. Three elements can be
noted which determine the effectiveness of the
information flow between a human and a
system: *
* The range and number of operator errors
in the formulation of queries to the system.
* The extent of the operator's lack of
knowledge of the system and commands.
* The extent to which the repertoire of
sequences of commands or queries occurs in
each working session, and therefore reduces
progress.
In this paper these aspects will be
discussed with regards GIS software, together
with some particulars on the implementation of
three procedures in an experimental GIS
software package, which give the system a
certain level of quasi-intelligence.
2. AN EXPERIMENTAL GIS.
In order to perform experiments with an
implementation of some quasi-intelligent
features in a GIS software, an experimental GIS
has been developed. The proposed GIS software
was created to the satisfy the following
assumptions:
(i) The database should be optimized with
regard to storage and accessibility of
information. The criteria for the optimization
would be based on the time needed for the
completion of a typical request.
(ii) The communication language between
GIS software and and operator should be in
terms of simple statements in English
language.
(iii) Every response of the GIS should have a
graphical form of display for visualization as
well as tabular or other forms of displays.
(iv) The GIS should be able to present all
possible relationships (both explicit and
implicit) between the primary data stored in the
database on request.
(v) The GIS database should be able to store
and process spatial information in both vector
and raster form as well as textual data.
(vi) The GIS should be equipped with quasi
intelligence.
The GIS with these features would be
described as user-oriented. Each of these
conditions will be disscused below.
(i) It has been assumed for this study that the
RDBMS ORACLE currently available on the
market is optimal with regard to condition 1.
(ii) The proposed GIS fulfils condition 2 by
the STRUCTURED QUERY LANGUAGE
(SQL). (ISO 1987). This language is of a very
high level, differing little from a natural