[Modeling] Class Diagram Specification

Marc-Philippe Huget M.P.Huget@csc.liv.ac.uk
Tue, 03 Jun 2003 17:52:27 +0100

"Wagner, G.R." wrote:

> Usually, we start with conceptual/domain modeling, where we use the
> attribute concept (with implementation aspects such as visibility)
> for modeling properterties.

That's not what I am doing, I will be found really weird  if I am doing
first: OK, I need a unique identifier for my agent, let's say GUID: String,
I suppose you define as well the type of attributes, isn't it? I prefer
without doubt OK, I need a unique identifier, [later] I have my agent
description, let's see how to define them, I can take a String to this
purpose. Moreover, maybe I am wrong that restraining too early how to
represent agents is a big problem and reduce the ability to use the same
agent in different languages, I prefer to have a nice agent then deciding
how to represent them, in modal logic, in UML, in Object Z or whatever than
an agent fully conformed to UML and a lot of pain to translate into
something else.

> A belief is always expressed in some language defined by a schema
> which may be expressed by a UML class model (or an RDF/OWL schema).
> So, in this example the belief refers to an instance of the object
> type City and attaches a special value ("sunny") to the attribute
> TodaysWeather.

Frankly speaking, I can't do that before fully describing my system.


Marc-Philippe Huget

Agent Applications, Research and Technology Group
Department of Computer Science
University of Liverpool
Chadwick Building, Peach Street
L69 7ZF Liverpool
United Kingdom

email: mph@csc.liv.ac.uk