Geoscience Reference
In-Depth Information
violated however the choice of actions is based on limited set of rules and
does not consider the quality of the result.
Lamy et al (1999) used more intelligent agents which can make an action
list and choose actions from it. They decomposed map objects into agents
at three different levels, macro agent, meso agent and micro agent (
Fig 3
).
Lower level inherits higher level's attributes. Decomposing objects parti-
tions the map into several parts, making the generalization process more
efficient.
Figure 3:
Decompose map into three levels of agent Lamy et al (1999)
The life cycle of a meso-agent when a constraint is violated and action is
needed is presented in
Figure 4
from Duchene (2001). Communication be-
tween meso agents and micro agents are described. Meso agent makes a
plan of its actions for generalization. Then it sends a signal to the micro
agents it contains to manage their own generalization. Each micro agent
makes its plans and evaluates the best one. Then meso agent re-evaluates
its situation. Once all evaluations are done, the meso agent chooses the
best plan for itself.
