[topicmapmail] reification and other funny things

[Lars Marius Garshol]

* Joril Andersen
>
> ....the data model 2.0 says that a topic has null or one  
> information item in its reified property.
>
> In the TMAPI the topic.getReified() method returns a set of topic  
> map constructs that this topic reifies.
> Which is correct?

The TMDM is correct. TMAPI has this wrong. A topic can't reify more  
than one thing, as this would violate the "one subject per topic"-rule.

> If a topic reifies an occurrence, I would think that this  
> occurrence can be assigned to several topics.

There are different ways I could interpret what you say here. Do you  
mean that multiple topics can reify the occurrence (this is not the  
case; they must in that case merge), or that the occurrence can be an  
occurrence of multiple topics (also incorrect, per TMDM), or  
something else?

> And hence the reified topic would contain a set with alle these  
> occurrence items in the different topics?

Now I'm lost. Above you were only talking about a single occurrence  
item, but now there are suddenly more of them. Which occurrence items  
do you mean?

> A document on well-known composers could be attached to both Puccini 
> (topic) and Verdi(topic)? Or should an occurrence be more  
> specificly related to a topic?

There is a difference between the document and the occurrences. The  
document is an information resource, but the occurrences are the  
relationships between various topics and the information resource. In  
other words, you can have (using LTM):

   {puccini, article-about, "http://example.com/italian-composers/"}
   {verdi, article-about, "http://example.com/italian-composers/"}

but this is two occurrences, not one, even if there is just one web  
page involved. So if you created a topic to reify the first  
occurrence that topic would not (and could not) reify the second.

> During deserialization, wouldn't the list of reified properties be  
> created for each topic?

Here I'm lost again, I'm afraid. What do you mean?

--Lars M.