[topicmapmail] Questions About XTM Occurrence Semantics

[Peter P. Jones]

Hi Steve,

Thanks for chipping in. Comments below...

Steve Pepper wrote:
> * Peter P. Jones
> | 
> | Assume there is an information resource on the web - an image in a
> | document. As a topic map builds up three topics come into being {A,
> | B,C}, each with different subject identifiers, but all of which refer to
> | that image from an occurrence. In each case the occurrence referring to
> | the image is in a different scope, but each occurrence has the same IRI
> | value.
> 
> There seem to be some communication problems here, which may be due to your
> use of the term 'occurrence' in two difference senses: the one formal and
> correct, the other colloquial and not quite accurate. Here's the definition
> of occurrence in [1]:
> 
>    3.9 occurrence: a representation of a relationship between a
>    subject and an information resource
> 
> Note: An occurrence is *not* the information resource itself. Therefore:
> 
> 1) An occurrence *can* be scoped, but it can *not* have an "IRI value"
> (unless you means an item identifier, about which more below).
> 
> 2) The *information resource* (sometimes colloquially, but inaccurately,
> referred to as the "occurrence") cannot be scoped, but it *can* have an "IRI
> value" (if it is external).

OK, good. That is better.
I was confused by:
5.6 Occurrence items
[...]
"Occurrence items represent occurrences. Occurrence items have the 
following properties:
1  [value]: A string. If the datatype is IRI, a locator referring to the 
information resource the occurrence connects
    with the subject; otherwise the string is the information resource."
In the latest public TMDM draft.

So presumably I should stick to the term 'occurrence /item/' [my 
emphasis] before referring to IRI values of 'occurrences' [my misuse].

> 
> Given 1) and 2), the situation you describe ("in each case the occurrence
> referring to the image is in a different scope, but each occurrence has the
> same IRI value") is not possible. (Even if we were to assume that you mean
> "item identifier" when you write "IRI value", the situation would still not
> be possible, because three different occurrences will always, by definition,
> have three different item identifiers.)
> 
> | Topic A : Scope D : Occurrence #1 : Value IRI V
> | Topic B : Scope E : Occurrence #2 : Value IRI V
> | Topic C : Scope F : Occurrence #3 : Value IRI V
> 
> I interpret this as follows:
> 
>          scope D
> (A) ---------------> V          .... #1
> 
>          scope E
> (B) ---------------> V          .... #2
> 
>          scope F
> (C) ---------------> V          .... #3
> 
> 
> Where (A), (B) and (C) are topics, each of which has an occurrence whose
> value is V, with the occurrences being in different scopes. Have I
> understood this correctly?

Yes, that looks closer to what I am after - the representation of the 
intent. I assume the #x numbers are just either comments or possible 
'item identifiers' that could be added in?

> 
> | I would then like to be able to reify one of those arcs.
> | That specific reference from a particular 'position' within the topic
> | map is what I would like to make the subject of a reifying topic.
> | I want to capture being able to talk about that particular arc between a
> | topic and the specific subject appearance indicated by the occurrence
> | value.
> 
> This is precisely what happens if you reify one of the occurrences. You get
> a new topic, as follows:
> 
>    (T) - a topic representing the (scoped) relationship between
>         (say) (A) and V
> 
> This topic can then be used to make assertions about the relationship in
> question.

Excellent.

> 
> | For me, it is less about having a topic that represents the arc so that
> | I can say in an association, "Jane inserted an information item with ID
> | 345 on May 4th, 2007," and more about having a topic that represents,
> | "This particular indication within the TM of the appearance of a subject
> | in the infosphere." Something that is as close as I can get to, "This
> | particular pointing out of a subject."
> | 
> | Another way of putting it is that I want to be able to look at the
> | reifying topic and know that it is talking about the 'particular
> | pointing out of a subject' and not just 'arc number 345 in the TM'.
> 
> If I have understood correctly, then this is exactly what you are able to
> do. The new topics represent the *relationships*, not the "arcs" (or their
> equivalent in correct TM terminology: there is no concept of an arc in later
> versions of 13250 that are not based on HyTime).

That is good. Taking my passage above, my intuition is still that there 
might be some distinction between my referring to the occurrence as 
construct in the TM model as opposed to its semantic intent [my 
terminology]. Your explanation suggests that the semantic intent part is 
covered.

If I refer to the item identifier of the occurrence item using a subject 
locator, and make that subject locator the subject identifier of a topic 
(reifying?) do I then get to attach say, metadata, to the construct 
purely qua construct? [my terminological spaghetti].


> 
> | Does that make sense?
> 
> Only just :-) It would help a lot if you could frame your question as much
> as possible using the terminology of the standard, otherwise it's hard to
> know what you mean.

I'm getting there; but so many things have been rejigged since I last 
looked at TMs... ;)

Sincerely,
Peter



> 
> Best regards,
> 
> Steve
> 
> [1] http://www.jtc1sc34.org/repository/0696.pdf
> 
> 
> 
> 
> 
>