Paul Sandoz wrote:
> It would not be necessary to do exact matches based on validation.
>
> All that would be necessary to do is given a set of qualified names (
> {namespace}localName ) in the schema count how many occurences of those
> qualified names occur in a set of n documents.
>
> Once everything has been counted sort the set of qualified names
> according to the number of occurences, the qualified name with highest
> number of occurences being first.
>
> Then assign an index to each qualified name whose value is position of
> the qualified name in the sorted set.
I see. Well, then do you really need a schema? Can't you just build up
histogram of names just by looking at some number of instances and
assign indices to them?
In any case, if you need XSOM help, please let me know...
--
Kohsuke Kawaguchi
Sun Microsystems kohsuke.kawaguchi_at_sun.com