Phonetics for OPC metadata

[MURATA Makoto]

Dear colleagues,

I am wondering how we can add phonetics for Japanese Kanji text
as part of OPC metadata.

There are four approaches:

1) Introduce a phonetics element for each of the metadata elements,
2) Introduce an attribute to each of the metadata elements
3) Allow the use of HTML ruby (and bidi) as part of the existing
metadata elements
4) Allow the use of Unicode ruby characters as part of the existing
metadata elements

ONIX has adopted 3).  Made-in-Japan metadata schemas (such as DC-NDL)
use 1).  Which do people here prefer?

Note: Neither Atom nor ODF is free from this problem.  Ideally, the
same solution should
be used everywhere.

Regards,
Makoto