[topicmapmail] graphic language for describing TopicMaps

[Are Gulbrandsen]

On 7. mai. 2004, at 19.23, Stefan Lischke wrote:
> But is there a standardised way of creating a digram which describes a 
> topicmap or a topicmap typesystem which does not only include topics, 
> it also includes association types with specified roles and 
> cardinalities.

Good question!  :)

> How are you guys describing your topicmaps? Are you using rectangles 
> for topics and circles for associations, how do you show that 
> something is scopeable?

I have been looking for a good graphical notation for Topic Maps 
ontology modelling. A graphical representation of a TM schema, - not 
the instances.

I had a discussion with Lars Marius Garshol on this, one and a half 
year ago, at the first Norwegian Topic Map conference. My opinion was 
that I thought ORM might be a good match for modelling Topic Maps, but 
he thought UML would be better. After this I started experimenting a 
little with different ways of visualising the ontology.

I'm used to conceptual data modelling in ORM (Object Role Modelling) 
(also refered to as NIAM), and think in some respects this modelling 
technique is closer to the TM model than UML. (For more info on ORM 
http://www.orm.net/ is a good source of information)

The article "Networked Knowledge Representation and Exchange using UML 
and RDF" is interesting. In chapter 2.1, "Ontologies as class diagrams: 
the family ontology" the author explains how to use UML to model 
ontologies:
http://jodi.ecs.soton.ac.uk/Articles/v01/i08/Cranefield/

I think neither UML or ORM is a perfect match for Topic Maps, so I have 
kind of developed my own diagram notation, mostly inspired by ORM. It's 
only a sketch, but this might be a good opportunity to see if anybody 
think it's interesting enough to comment on it.

Compared to Figure 1 in the mentioned article, I think my "TORM" 
diagram is much easier to read. (Maybe not surprising since I made it, 
but a more interesting question is what you think.):
http://folk.uio.no/areg/topicmaps/graphicalTM.html

It's getting a bit too late, but I think the diagrams should express 
roughly the same ontology. (Apart from errors and misunderstandings, I 
don't think the author and I agree on all details of how to model a 
parent-child association, but I might change my view after getting a 
few hours of sleep... :)


Best Regards,
Are D. Gulbrandsen
The XML-group,
Center for Information Technology Services
University of Oslo