[sc34wg3] Reification or Representation?

[Kal Ahmed]

On Wed, 2004-08-25 at 06:11, Steven R. Newcomb wrote:

Please excuse the edits:

> (1) representation: I agree with Murray.  "Representation", for me, is
>     symbolization.  Symbols mean what they mean regardless of their
>     inherent properties, and their inherent properties need not
>     correspond to, or be analogous to, the properties of whatever it
>     is that they symbolize.
<snip/>
> (2) proxification: This is the same as representation, in the sense
>     that the proxy of a subject can (and should) be regarded as a
>     symbol for that subject.  But it's not just representation; it's
>     more than that.  The proxy of a subject has at least some of the
>     same properties (or attributes, or whatever -- let's have that
>     argument separately, please) as the subject itself.  Or, rather,
>     the proxy has properties that are somehow analogous to some of the
>     properties of the actual subject, so that the proxy can be used to
>     guide certain kinds of manipulations in a way that, if it were
>     possible, the subject itself would presumably guide in an
>     analogous fashion.
> 
>     If we see topic maps as sets of subject proxies -- as graphs of
>     nodes whose asserted relationships to each other are analogous to
>     someone's opinions about how the subjects of those proxies are
>     related to each other -- then such views of topic maps are clearly
>     cases of proxification, as opposed to mere representation.  The
>     reason I say this is that the properties of the proxies --
>     including the properties that are connections to other proxies --
>     are analogous to the properties of the actual subjects.
> 
The way I see it, only topic information items in the TMDM perform
proxification ("proxify" ? ;-). All of the other constructs in the TMDM
provide context for the proxified subjects.

> (3) bernersleeification: The operational definition of
>     bernersleeification depends on whether you regard a topic as a
>     mere representation of a subject, or as a proxification thereof.
>     If you're only concerned with representation, then it doesn't
>     matter if the result of bernersleeification is two or more
>     representatives for the same subject (that subject being the
>     relationship represented by the association that you've created a
>     topic for in order to bernersleeify it).  It's OK, for example, to
>     have multiple information items, one a TMDM association and the
>     other a TMDM topic, that each represent that relationship, and
>     that have special connections to each other.
> 
>     On the other hand, if you're trying to do proxification, the
>     result of bernersleeification needs to be a single proxy.  If it's
>     not a single proxy, then, if nothing else, there are either
>     spurious connections between the multiple proxies of the same
>     subject (connections that I'm calling "spurious" because they are
>     not analogous to relationships between subjects).  (But *having*
>     such spurious non-relationships between two or more proxies for
>     the same subject is whole lot better than *not* having them.  The
>     latter case, where there are multiple unconnected proxies for the
>     same subject, is analogous to "subject schizophrenia", because
>     that single subject appears to have different personalities,
>     depending on which of its proxies you happen to be using as its
>     proxy.)
> 
Following on from the position that only topic information items perform
proxification, I would say that association information items provide
context for the subjects proxified by the association role players. If
there is another topic information item that has a special relationship
to the association information item, then it proxifies the subject of
the relationship between the role players in the association information
item.

Under this view there still remains only a single proxy for the subject
of the relationship - the association by itself does not proxify the
relationship.

I would argue that the association information item cannot proxify the
subject of the relationship because it does not allow us to place the
relationship in any other context (we cannot attach names or occurrences
to it, nor can we make it play roles in other associations). There is a
weak proxification because the association information item does provide
for type and scope, but I see that as part of the context provided to
the topic information items by the association information item.

Cheers,

Kal

-- 
Kal Ahmed <kal@techquila.com>
techquila