detection, -maximum intensity for sturation, — 
maximum storage, —delay for acquisition, -delay for 
transmission, — -delay for processing, -a 
BEHAVIOR, which is an algorithmic part, 
composed of any set of executable statements, this 
behavior being executed when a launcher is 
modified. The figure 7 shows the graphical 
representation of the neurons. A "launcher" may be 
any attribute or link, which, when modified, will 
launch the behavior. 
  
min.intens.for detection 
max.intens.for saturation 
    
   
maximum storage 
delay for acquisition 
delay for transmission 
J delay for processing 
et ur (launcher) 
p FE 
Behavior 
  
     
  
  
  
3 
C» C»C»C» (interactions) 
  
  
  
Fig.7 : Hyperclass NEURON and its components. 
For instance, on the figure 8, we see a neuron of the 
class FIRE HYDRANT which launcher is modified 
by an external intervention, which behavior consists 
of modifying the pressure of another neuron of the 
class SECONDARY PIPE, and the behavior of this 
last one modifies the state of another neuron of the 
class surpressor, etc... 
  
external 
intervention 
    
  
   
   
  
  
   
   
    
     
   
  
  
m diameter 
oO delivery 
C pressure 
D state 
FIRE HYDRANT 
SECONDARY 
PIPE 
pressure P 
O input pressure 
CInominal output pressure 
Cloutput pressure 
behavior 
“behavior 
  
behavior 
etc... 
  
  
Fig. 8 : Neuron interaction and chaining. 
  
42 
Neurons too may be fuzzy (Bouille, 1991b); they 
are very usefull in terrain modelling (Bouille, 1992). 
Rules, processes and neurons have not exactly the 
same goals; though anyone may sometimes replace 
another one (Bouille, 1993a, 1993c), they are more 
dedicated to specific topics, respectively expert 
system, simulation, automated learning. 
Nevertheless, they may interfer, a rule possibly 
modifying a fact which is a launcher for a neuron, 
the behavior of this one then activating a process 
which body will modify another a.d.t., as 
represented on the figure 9. 
  
E 
     
     
Facts 
] 
NEURON auncher 
Behavior 
  
Ts 
© Process D 
il Sire 
etc... 
  
Fig. 9 : Mixing rules, processes and neurons. 
All the preceding tools are required for the present 
geographical applications; in fact, this requirement is 
not new  (Bouillé, 1981) we would like to 
emphasize the need of a graphical representation of 
all types previously mentionned, together with a 
programming language based upon a perfectly 
defined grammar; that is the purpose of the 
programming language ADT'81, particularly 
convenient to geography and cartography (Bouillé, 
1994a). Many applications have been developped; 
among others, digital and topological terrain models 
are very promising (Bouillé, 1987), (Baton-Hubert, 
1994), (Hubert, 1991, 1993a,b) as well as 
automated positionning of the toponymy on the 
maps (Titeux, 1989). 
The complete kernel of the HBDS system includes 
several tools interconnected, we just here mention: 
-a multi-level indexed sequential file system, 
-a very large object-oriented database,based upon 
some decisive criteria (Bouillé, 19912), 
-a multi-engine fuzzy expert system managing an 
illimited set of persistent rules and facts,the database 
thus becoming a very large knowledge base, 
—a simulation system with an engine managing an 
  
illimited s 
second er 
—a neura 
object-or 
an auton 
expert sy: 
—a 3D ste 
multimed 
-a specif 
1986), ir 
robotics t 
decade. 
The comp 
an ADT'8 
an execul 
integrated 
(Bouillé, 
3. 
Any type 
looks like 
time, and 
is invoke 
concerns : 
a class, i 
following 
result gray 
CREATE " 
  
Fis 
  
Fig. 1 
In this fir 
concerned 
consider a 
be greater