<- PREVIOUS Chapter
T4. Design and implementation of Gestore Preferenze, GP
For Gestore Preferenze the domain of its activity is the Preference Base of Consigliere Diretto
The PB uses the same class concepts as MKB and OKB. This concepts can be interpreted as objects in the
SDB of GP, it is called PSB (Preferences Situation Base). The relations between such objects is defined by preference relation.
The Fig.6. illustrates a possible graphical Object-Based vision the PDB of the Consigliere Diretto in the PSB, from the perspectve of Gestore Preferenze.
In the functional perspective, Gestore Preferenze can be seen as a manager/supervisor of Consigliere Diretto which, for the situations of the lack of success of his worker, has adequate meta-preferences and adequate knowledge according to which he can change the preferences of the worker.
He use meta-concepts defined from the point of view of the emergency-management goal. He is goal-oriented, he bases on qualitative evaluation of risk related to the some class of objects (i.e. = states of the emergency domain). He needs to know safety criteria related to the state of the emergency domain.
T5. Design and implementation of Pianificatore
By analogy to human relations, Pianificatore has a role of an methodology expert. He can see MKB and OKB of Consigliere Diretto, knowing also current intervention goal he shoud be able to preper plans or to deduce new knowledge for CD. In this sense Multi-agente has ability to learn.
For example, if the CD has two methods:
M1(current_state -> X_state)
M2(X_state -> goal_state)
then it is not able to known that the goal_state is possible to achieve.
In such case Pianificatore must intervent.
The Fig.7. illustrates a possible graphical Object-Based vision of the OKB of the Consigliere Diretto in the KSB (Knowledge Situation Base), from the perspectve of Pianificatore. KSB is a basic domain of activity of Pianificatore.
Some interesting data about possible Planning Agent: http://
Other potentially useful references will be distributed during partners meeting 6/6/97.
T6. . . .
5. Final Comments
We would like to stress that the multi-monad architecture of simple personoid, for Multi-agente here used, is currently the most advanced theoretical approach in this field being developed in ENEA. It can be modified, extended , or... abandoned . This last possibility will be taken into consideration if any other generic multi-agent repetitive architecture, on at least the same level of theoretical development, is proposed.
<- go back