[sc34wg3] What do we mean by reification?

[Patrick Durusau]

Lars,

Lars Marius Garshol wrote:

>* Patrick Durusau
>|
>  
>
<snip>

>  
>
>| Speaking solely for myself, I have no objection to "representation"
>| but I would use in all cases. If it is representation in one, then
>| it seems to me it is representation in the other. (or the converse
>| as well, reification in one is reification in the other, but as you
>| point out, correctly in my opinion, it requires more explaination
>| and hence "representation" should be preferred.)
>
>In the RM this works, but in the SAM it does not. SAM needs a term for
>the trick used to make topics represent base names/occurrences/assocs,
>but the RM has no use for it.
>  
>
I think the part that may be separating us is the following line in the 
SAM (emphasis added):

*One may want to give an association occurrences, or to give an 
occurrence a name. **The basic topic map model does not allow this**, 
but through reification this can be done by creating a topic that 
reifies the topic map construct."

Where I am lost here is the **The basic topic map model does not allow 
this** portion. If a subject is anything whatsoever I want to talk 
about, why are association occurrences and giving an occurrence a name 
excluded from that ambit? (Is this a question of XTM syntax? That may be 
where you are losing me.)

In that case, is the SAM tied to XTM syntax? (Not an objection, just a 
question.)

<snip>

>| So, if we agree that the term "representation" is better for
>| describing the topic / subject relationship in terms of clarity (I
>| assume to people who are not part of the topic map community) what
>| is being gained for implementers by using a special term,
>| reification, consistently with the AI folks? Not that I particularly
>| mind being consistent with the AI crowd but are we gaining anything
>| by that consistency?
>
>Why are you talking about implementors?
>
>What we gain is having a term for the trick used to make topics
>represent topic characteristics. I think a term for that is necessary,
>but if people disagree we should discuss that.
>  
>
And how does the name help? (This is another place where I am lost.) If 
I follow the rules in the SAM, whether I call it "representation" or 
"reification" isn't the use of a subject identifier going to operate the 
same way in terms of being detected?

>| I think what I am missing is what makes this a "special case" of
>| topic-represents-subject senario. At least in terms of why I should
>| say reification versus representation.  Noting that if a subject
>| identifier property of a topic equals the source locator item of an
>| information item, then that topic reifies (in the current SAM sense)
>| the information item. Admittedly mechanics are different from a
>| topic that "represents" (in the SAM sense) a subject not in the
>| topic map, but does the difference in mechanics merit a separate
>| term for the relationship between topic / subject?
>
>I think it does. Imagine trying to rewrite the SAM without this, and
>trying to teach people XTM without being able to use this (or some
>other, more appropriate) term for this concept.
>  
>
Well, probably doing a lot of violence to the reification section of the 
SAM but I did re-write that section without using reification, which may 
or may not be indicative of the need for a separate term:

*******WARNING***This is merely an example of a suggested usage of 
language in the SAM. It has no connection with any past, current or 
future versions of the SAM***

Every topic represents one subject, and the relationship between the two 
is always one of representation.

In many cases it is desirable to be able to attach additional 
information to topic map constructs such as topic names or associations. 
One may want to give an association occurrences, or to give an 
occurrence a name. The basic topic map model does not allow this, but 
this can be done by creating a topic that represents the topic map 
construct. The necessary information can then be attached to that topic, 
and the represented relationship is present in structured form, and can 
reliably be detected by software.

This representation is achieved by giving the topic a subject identifier 
that refers to the topic map construct that is being represented. In 
model terms, this means that if an information item has a source locator 
item that is equal to one of the items in the [subject identifiers] 
property of a topic, that topic item represents the information item.

Note that one topic cannot represent another. To make one topic the 
subject indicator of another implies that the two topics represent the 
same subject, and they will therefore be merged, and thus become a 
single topic.

***************End example of possible usage******************************

Hope the formatting works better this time!

Patrick

-- 
Patrick Durusau
Director of Research and Development
Society of Biblical Literature
pdurusau@emory.edu
Co-Editor, ISO Reference Model for Topic Maps