[osis-core] Index Syntax

[Patrick Durusau]

Harry,

Harry Plantinga wrote:

>>Yes; if the index entries are being marked, hopefully they are being 
>>marked in the text where they go; in which case you can either 
>>auto-generate a page or canonical id as appropriate, or do without 
>>(since the index is essentially the result of a query for all these 
>>index-entry elements, and in the result, you could have anything you 
>>wanted, since the machine can take you right to where it was 
>>generated from.
>>
>>...
>>
>>Oh, one other nit: Should we do anything for 'see' and 'see also'?
>>
>
>I've been hesitating to bring it up, so thanks!  There probably should
>be a way of handling see and see also.  There could be several 
>'see also's so a single seealso attribute is not adequate. How does
>TEI handle these issues?
>
Not with the <index> element or its attributes (as far as I can tell). 
The attributes are level1 - level4 and appear to correspond to what I 
have suggested as key1 - key4, since they are not really levels of indexing.

While I see the value of the see/see also (very similar to the 
declaration of relationships in topic maps) I am not sure that such a 
mechanism can be easily implemented as part of the core schema.

For example:

<index index-name="subject" key1="Job" key2="theodicy" key3="parallel 
literature" key4="Babylonian sufferer">

In some sense you could already extract see or see also from the 
existence of multiple keys, that is why they all occur on the same index 
element I would assume. Not sure I would have completely unrelated keys 
on a single index element.

But that relationship is merely implied and not guaranteed to be a 
useful one.

On the other hand, given the foregoing syntax, how do I declare a "see 
also" pointer for "theodicy" and not any of the other attribute values? 
I can imagine a header (getting real crowded up there!) mechanism that 
declares "see" and "see also" values, perhaps as:

(possible proposal, not in syntax presently)

<blort id="121212">theodicy
    <relatedBlort type="see">prolem of evil</relatedBlort>
</blort>

And the key attribute merely has IDREF to the <blort>.

But, do we need such a mechanism for basic markup in the core? Perhaps, 
I am willing to do it if we can quickly settle on a syntax and get the 
remaining issues resolved fairly quickly.

Patrick

>
>-Harry
>

-- 
Patrick Durusau
Director of Research and Development
Society of Biblical Literature
pdurusau@emory.edu