(iii) Usually a GIS is used by users to perform 
typical tasks which are a sequence of the 
queries. The difference between various tasks is 
the data which the system operates on. The 
introduction of the same sequence of queries 
every time is time consuming and therefore 
inefficient. 
In the following we discuss the solutions, 
which improve the effectiveness of operation of 
the GIS through the elimination of the above 
mentioned sources of inefficiency. 
The procedures have been implemented 
in the experimental GIS described above. 
3.1 Procedure for Identification and 
Correction of Errors. 
The aim of the procedure is the correction 
of the syntax and the semantic errors, which can 
occur in queries formulated in SQL for GIS. In 
particulary the procedure works as follows: 
(i) A query is analysed by standard rules on 
ORACLE RDBMS in order to detect the syntax 
and semantic errors. 
(ii) If no errors are determined, then this type 
of query is stored in a special database. In this 
database are stored all queries which have been 
used during all work sessions of the system. If 
this type of query does not exist, the procedure 
will add to the database. If the query is found 
then the procedure takes no further action. 
(iii) If an error is detected, the procedure 
searches for the most similar query in the query 
database. The located query is then executed 
with actual values of parameters, after its 
acceptance by the operator. 
(iv) The procedure introduces into a special 
database which is created by the system for each 
particular user, information on the user's 
errors. The information is utilised to create a 
profile of the characteristic of the user 
performance on the system. This point will be 
discussed later. 
The most important part of this procedure 
is the algorithm for selection of the query which 
is most similar to that which has been introduced 
with error. The utilised algorithm must 
determine the most appropiate query on the basis 
of the similarity of each language element and 
its sequence. 
The effectiveness of the procedure 
depends also on the richness of the database of 
query patterns. Because of this, the query 
database is continually updated during the 
utilisation of the system, the number of wrongly 
framed queries will be decreased with the use of 
the system. The simulations have shown that the 
procedure works correctly in almost all cases. 
Wrongly selected queries have been observed in 
about 2% cases which occurred with badly 
deformed queries which are abnormal in 
practice. Thus, they should be given less 
emphasis in the evaluation of the procedure. 
3.2 A Quasi-intelligent "Help". 
The amount of information which is 
provided by an on-line "help" system to the 
operator is usually much more then each 
operator needs. This redundancy causes 
problems with the recognition of particulars 
which are needed. It is much more difficult to 
find information if the volume of the message 
requires more then one screen for display. 
Therefore it would be much more convenient if 
the system supplied only the information which 
the user needs. The problem can be addressed by 
giving a value to each user which will be an 
evaluation of the user’s needs. This value will 
determine how much information will be 
provided with help. The value of the user's needs 
can be made on the basis of the following 
quantitative characteristics of the user's work 
with the system: 
(i) Frequency of use of each type of query, 
(ii) Frequency of errors made with each type 
of query. 
In this implementation of the quasi- 
intelligent "help" the evaluation of the user's 
needs has been made using the second method. 
In particular the implementation can be 
introduced as follows: 
(a) For each user the system creates a 
database which we call "The characteristic of the 
user". In the database are stored the information 
about the amount of use of each type of query and 
the number of the errors which have occurred 
using each query. This information is provided 
by the procedure for identification and 
correction of errors (Figure 1). 
(b) On the basis of the stored information the 
frequency of the errors (p), which occurred for 
each particular query is calculated. This value 
determines which level of the help message will 
be displayed to the user.