Michael Beddow's interventions in the the "index" thread, in particular
his remarks about cross-references from one index entry to another within
the same index, got me wondering how markup for this may have been
achieved in P4.  the following examples may or may not be germane to the
questions raised about the proposals under P5. They may serve to elaborate
the typology of "indexical relations" so ably sketched out by Michael.

Example:

<index corresp="id02" ana="bicr" id="id01" index="works" level1="Works and
Days"/>

<index corresp="id01" ana="bicr" id="id02" index="authors"
level1="Thesiod"/>

<interp id="bicr" value="bi-directional_cross-reference"/>

<index copyOf="id01"/>

From this example, one could generate two different types of
cross-references between two different indices:

        The redirect e.g. See...
        The additional direction e.g. See also...

Bonus: such mark-up would also facilitate processing to create a combined
index of works and authors to be generated.  The "copyOf" attribute could
be used to control the placement of other <index> elements with the same
content. I stress this point because I want to indicate that the relation
between <index> elements marked up here with the "corresp" attribute is
not a relation between the places in a document instance where <index>
elements are located. For example, it is possible that the name of an
author required for the generation of an index entry is not present at the
location that one might wish to pick up for the index. In short, index
generation is about where, what and by what name.
where -- the location
what -- the index term, the index it belongs to, the level it belongs to
what name -- other terms that can serve to build a cross-reference to a
given index term

From a text-analysis perspective one could process a marked-up document
instance to ascertain what types of index terms cluster around which
locations. Somewhat like have a "reverse-index".

I gave some thought to the mark-up of mono-directional cross-reference. I
have not yet found a way to mark-up <interp> and <index> elements to allow
for a transformation to consistently generate the content of a <link>
element appropriately.

Example:

<interp id="monocr" value="mono-dir_cross-reference"/>

<index corresp="id04" ana="monocr abc" id="id03" index="works"
level1="History of Collage"/>

<index corresp="id03" ana="monocr abc" id="id04" index="authors"
level1="Wolfram, Eddie"/>

<link id="abc" targType="index" targOrder="Y" targets="id03 id04"/>


If cross-referencing is centred at the heart of index creation, 3-way
relations may be the way to conceptualize the problem. Michael described
"cross references internal to the index term listing". I think that is an
accurate way to describe the appearance of a listing of index terms.
Turning to P5, is there a need for mark-up of <indexTerm> elements that
can be linked via (some mechanism) other <indexTerm> elements? Do these
hypothesized <indexTerm> elements represent a significantly different type
that they deserve to be marked up under some element name of of their own,
say <indexAlt>? Might be useful for multilingual indices, etc.

And P5 offers the joys of recursive nesting...

<index>
<indexName>...</indexName>
<indexTerm> ...</indexTerm>
<indexAlt>...</indexAlt>
        <index>
        <indexName>...</indexName>
        <indexTerm> ...</indexTerm>
        <indexAlt>...</indexAlt>
                <index>
                <indexName>...</indexName>
                <indexTerm> ...</indexTerm>
                <indexAlt>...</indexAlt>
                </index>
        </index>
</index>

Interesting... would the content of <indexName> could be inherited?

If your head doesn't hurt yet... Would it be appropriate to mark up the
one instance of the location of an <index> element as a copy of itself and
all other location instances as copies of it

<index id="mb" copyOf="md"/>
<index copyOf="md"/>
<index copyOf="md"/>
<index copyOf="md"/>

Indexing as clone creation of cross-references.

--
Francois Lachance, Scholar-at-large
http://www.chass.utoronto.ca/~lachance/jardin

Skill may be the capacity to manipulate perceptions of knowledge.
Magic is.