[sc34wg3] SAM-issue term-subject-identity

[Ann M Wrightson]

It looks like we're moving towards (trying out the "subject identifiers"
terminology):

In a *given* topic map, (a set of?) subject identifiers *is* the identity of
the subject of the corresponding topic. (Do with it what you like, that's
what this topic map says, so there...)

When a tm-agent (human or machine), generates a new topic map from two input
topic maps, it uses the evidence supplied by the subject identifiers of the
topics in the two input topic maps (identity as that-identified), to
determine what topics are in the new topic map (identity as not-distinct).
(NB I'm not yet calling this a merge)

In the SAM, criteria are provided to which such an agent must conform if
this operation is to be a topic map *merge*. These criteria provide an
operational, SAM-level definition of subject identity (in both senses); what
you might call the subject identity policy of the tm-merging agent.
(It's then a question of SAM/RM mapping, to verify that this definition
satisfies an RM-level criterion eg as I gave in a  previous mail.)

And of course the TNC debate is one (important) aspect of the broader
question: What exactly should the subject identity policy of a tm-merging
agent be?

Ann W.