Class Discriminator - UML Best Practice

[Rouven]

Hello everyone,

I am looking for some input regarding a UML domain model. We have chosen a UML class model to represent our business domains. Now we encounter situations such as:
Class A abstract with attribute a
Class B is specialization of A with attribute b
Class C is specialization of A with attribute c

As a Java programmer a concept like "instanceof" comes naturally to me, but talking to business users this is obviously a problem. Does anybody have a best practice on chosing a model representation that allows the business users to describe that they have a list of object of type A, but only wish to process those that are of type B?

• introduce an attribute "type" as a discriminator? if so, how do you model it (static, read only, ...)?
• introduce the concept "instanceof"?
• other ideas?

Thanks & best regards,
Rouven

[Geert Bellekens]

I would simple write something like:

Code: [Select]

IF variable IS AIn an actual context that could be something like:

Code: [Select]

IF currentDocument IS Invoice
or a variation to do

Code: [Select]

FOR EACH Invoice in DocumentsToTreat
I would of course be even better to avoid any possible ambiguity by using the concrete subtype as much as possible.

I believe that would be clear for both business users as technical users, without the need to define "difficult" concepts such as "type" or "instanceof"

Geert