A few days ago, I was looking for an easy way to model XML-documents with UML.
For applications that are all about distributed documents, UML is probably irrelevant. So why UML? Well, I am working with people who like UML a lot right now
.
But even if I wasn’t, I would consider this: real documents live in a world full of other applications – and today most people model business applications in UML (if they model them at all
.
Also, if you build objectoriented applications using modern tools, implementation might become easier if everything is modeled as objects.
And integrating XML-editors with “ordinary” applications is probably easier if you think of documenttypes (and the concepts necessary to describe them) as just another kind of business objects.
OK. So if UML is a good idea, there are two basic problems:
- Choosing an XML-oriented syntax for document types
- Mapping it to UML
Picking a decent XML-syntax for documenttypes is the easy part. The W3C standard isn’t frozen – but any decent subset of the current standard should be easy to upgrade. Right now, I think Microsofts version (the one in IE5) is the best bet. What’s the world coming to!
UML-mapping is harder to say anything general about. I think it all depends on specific tools and applications. We’ll see. I have one case coming up in the next few days.



