[sc34wg3] revised draft Reference Model document N0298

[H. Holger Rath]

Bernard Vatant wrote:
...sniped a lot of interesting stuff...
> *Holger
> > If "graph" is understood as a collection of nodes
> > you can reach following the arcs starting at an arbitrary node ...
> 
> I think, Holger, you are wrong here. 

I somehow expected that ;-) because I should be but I am not very familiar
with graph theory (BTW: could you recommend a good book?).

> Your definition reduces the general notion of graph
> to "connected graph" (in which there exists a finite walk from any node to any other
> node). 

OK. I thought a graph has to be somehow connected. If it has not, even better.
Then the definitions in dRM make sense.

> I don't see why a topic map should be constrained to be connected. If you merge two
> topic maps with disjoint sets of subjects, role and association types, you will get a
> merged graph with (at least) two disjoint connected components. So being connected is not
> a requirement for G.
> 
> > ... and "subgraph" is understood as a subset of this
> > collection, then I would say that "every assertion within
> > a topic map" is not necessarily a subgraph.
> > Why? Because assertions may define graphs which are not
> > connected to other assertions/graphs.
> 
> I don't see your point. 

You cannot see the point because I was wrong. Sorry for the confusion.

> Two assertions A1 and A2 can be two disjoint subgraphs of G. This
> is not a problem in graph theory. Or an assertion could be a non-connected subgraph, if it
> is the union of A1 and A2, like "Bernard works for Mondeca and Holger likes Scottish
> beer" - although it would be quite easy to add a third assertion making the connection
> happen in that case :))
> 
> To be continued

Looking forward to reading more.

Regards,
--Holger