Re: [topicmaps-comment] Re: RDF/Topic Maps: late/lazy reification vs.early/preemptive reification

[Lars Marius Garshol <larsga@garshol.priv.no>]

* Lars Marius Garshol
|
| I think SRN is right here: the way to model this is to use a ternary
| association, with the sponsor playing the third role.

* Piotr Kaminski
| 
| Perhaps.  

I see no perhaps about it. I'm convinced that this is the way to model
it, and nobody has been able to suggest a better solution.

| This leads to another usage question: how do you decide when it's
| better to have a single association with multiple players for one
| role, and when it's better to have a bunch of n-ary associations
| (for fixed n, usually 2), with each role played by only one player?

I think in many cases one will see a need to capture a relationship
between two topics, and also a need to attach more information to
it. If that extra information takes the form of topics which can
reasonably be said to participate in the association on the same level
as the two first this is a candidate for a new role.

In other cases there will be many separate binary associations that
one realizes should be a single n-ary association. One example might
be the parents-child relationship, which has mother, father, and child
roles. 
 
| In any case, I think the problem is still valid; it's conceivable
| that I want to superimpose the sponsorship information on a foreign
| topic map without modifying it (e.g. so I can take advantage of
| future updates thereof).  If the original map didn't consider
| sponsorship, then the single collector approach would've been a
| valid (and, from what I've seen, preferred) one.

I agree that it is conceivable that you might want to reify the
association so that you could create another association in your topic
map to record this. But what is the problem?

By the way: please stick to topic map terminology on this list. We
have more than enough terminology confusion as it is. (Your
terminology may be good, for all I know, but it's unknown on this
list.)
 
* Lars Marius Garshol
|
| In PMTM4 there is no t-node that represents each association role
| (please correct me if I am wrong). This is in my opinion a serious
| weakness of PMTM4, and it's related to the problem of the
| three-legged arc.
 
* Piotr Kaminski
|
| Aha!  That's what I thought, but I wanted to take some more time to
| digest the heavy prose.  Has this criticism been made before, and
| was there a response?

I haven't seen it made before. In fact, I didn't realize this before
now. This is clearly one of the things we will have to iron out in the
process of perfecting and aligning the infoset model and PMTM4.

--Lars M.