[Modeling] Class Diagram Specification

Wagner, G.R. G.R.Wagner@tm.tue.nl
Tue, 3 Jun 2003 18:07:33 +0200

> crudely speaking, I would like to say that we have two levels: 
> agent level where we handle beliefs, knowledge and an "object" 
> level where we write the translation of this agent level into 
> something that can be depicted as a class diagram. Does it help?

I don't think so.
Obviously, a software agent (like other entities) has properties 
that are represented by means of attributes (e.g. its name, its 
GUID, etc.). In addition, it has beliefs (and possibly further
mental state components). As the name "mental state component"
suggests, its belief types may be modeled as (<<belief>> stereotyped) 
object types and other agent types, which are attached to the agent 
type under consideration by means of composition. They are 
containers for beliefs about objects and agents in the environment 
of this agent.

I've posted similar remarks already in response to a message from 
Bernhard Bauer (dated 15.4.2003).

Yes, agents are different from (non-agentive) objects. But since
they are also entities with properties and other features (such as
mental state components), we may well use the UML concept of a type
and the general notion of a "class" (in the sense of a type,
not in the sense of an <<implementationClass>>) to model agents.


Gerd Wagner  
Dep. Information & Technology 
Eindhoven University of Technology  
Email: G.Wagner@tm.tue.nl 
Phone: (+31 40) 247 26 17  
Fax: (+31 40) 247 26 12