[Modeling] AUML diagrams
Tue, 15 Apr 2003 18:02:03 +0200
> > Obviously, the beliefs of an agent can be most naturally
> > specified by indicating the (agent and object) classes
> > which belong to the agent's state structure definition;
> > so the agent has beliefs about (other) agents and objects
> > of those types. In general, there will be a number of
> > (agent and object) classes which form the basis for
> > shared beliefs of several agents, but there will be
> > also other classes that define private beliefs (in
> > AORML they are called "internal object types").
> I.e. the beliefs of agent are usual instances of some classes, right?
Yes, if there is an instance with ID=052164819X in the current
extension of the Book class of the library assistant (software agent),
then this agent believes that there is a book with ISBN=052164819X,
whose title is "Advanced Object-Oriented Analysis and Design Using UML"
and whose author is Jim. This corresponds to the following wff:
(exists x:Book)(x.ISBN=052164819X & x.title="Advanced Object-..."
& x.author=" James J. Odell")
Each instance of a class belonging to the belief state (or knowledge
base) definition of an agent represents such a wff/sentence.
Only more complicated (e.g. disjunctive) beliefs cannot be expressed
in this way.