language, and with simple grammar. This 
language was developed especially for the 
purpose of communication between the database 
and user. 
(iii) The sense of sight conveys 90% of all 
information absorbed by a human. Information 
contained in a picture is not transformed by a 
sequential process. This phenomenon enables 
abstract thinking processes, which are the main 
part of creative processes needed for planning 
etc. Therefore it is important to equip the user- 
oriented GIS with a graphical visualization 
system (Steiner et al. 1989). This system should 
present each GIS answer in adequate graphical 
form. 
(iv) The information gathered in a GIS is of 
quantitative and qualitative character. 
Quantitative elements are those in which the 
relation of order can be defined, while 
qualitative elements are all others. Both sets of 
information are described as primary or basic 
information. For each element of each set of 
quantitative and qualitative data, a relation can 
be determined with all other elements of the data 
sets. This will generate a huge number of 
relations, but only a small number of them have 
sensible meanings. The fourth condition can be 
fulfilled in 2 ways: 
* through the requests formulated in SQL 
or 
* through questions for which the answer 
would come as a result of operations on bitmaps. 
The second solution is much more 
effective than the traditional one ( i.e. 
analytical ) because it is based on the utilisation 
of fast operations on binary sets. The binary sets 
are created through producing graphic outputs of 
the GIS. 
(v) The fulfilment of the fifth condition is 
possible by utilizing a long data format 
available in RDBMS ORACLE records. Long 
data format enables the storage of 65535 
characters in each record. This means that the 
database can store objects such as image data in 
raster form. 
(vi) The software packages available on the 
market so far operate in such a way that they 
react directly to each request, either 
communicating to the user that an error has been 
made or performing a requested task. The 
quasi-intelligent system however would include 
in its operations not only the information given 
directly by the user, but also its own knowledge 
gathered about each particular user by observing 
that user's actions. That means that a quasi- 
intelligent system should be able to perform the 
operations of observing and gathering 
information about each user, and storing and 
making decisions on this information. 
The theoretical considerations lead to the 
conclusion that there is a distinct possibility of 
implementing quasi-intelligence into GIS 
along those lines. 
The two main fields in which the quasi 
intelligence would be useful are the control of 
proper use of the system and help given to the 
user. For instance, if the operator makes a 
syntax error, the system should make an effort to 
correct this mistake. If the operator for the first 
time decides to use some options available in the 
package, the system should inform the operator 
about its application and give him all possible 
assistance. The amount of assistance supplied 
by the system itself should be determined by the 
level of experience of each particular operator, 
which can be measured for example, by the 
number of syntax errors made or the time used 
by the operator during working on various 
tasks. 
The experimental GIS developed at the 
UNSW, School of Surveying has been 
constructed using of ORACLE RDBMS and other 
associated products such as SQL*PLUS, 
SQL*LOADER, PRO*C. The graphic 
subroutines have been written in C procedural 
language and using standard graphic packages 
GMR and GPR which are provided with the 
APOLLO DN 4000 computer. 
Interesting applications of ORACLE 
RDBMS have been presented in (Abel 1989 and 
Pong-Chai Goh 1989). 
3 QUASI-INTELLIGENT PROCEDURES. 
There are three main sources of 
inefficiency in GIS software: 
(i) The Structured Query Language has its 
own syntax and semantics. A query which is not 
formulated correctly according to the rules will 
result in an error message, after which the 
operator must, in many cases, re-enter the query 
from the beginning; this is time consuming. 
(ii) The use of complicated systems needs 
extensive training. The best form of the training 
is working with the system. Therefore the 
system should be equipped with a "help" or self 
training unit. This option should be started 
automatically, if required, in the context of the 
introduced query.