Need help with Oracle Text

Question

Hello,
I have the following table:
create table table1(col1 clob)
In columns COL1 I store some XML with the following structure:
<ROW>
<NAME>....</NAME>
<COL1><VAL>...</VAL><VAL>....</VAL></COL1>
<COL2>....</COL2>
</ROW>
So I compare two indexes:
1. Simple index:
create index myindex on table1(col1)
indextype is ctxsys.context
parameters ('datastore ctxsys.default_datastore
filter ctxsys.null_filter
section group ctxsys.auto_section_group')
2. Another index with stop sections, I think it will allow to increase performance and decrease index's size as I will index only NAME tag.
begin
     ctx_ddl.create_section_group('xmldatagroup', 'AUTO_SECTION_GROUP');
     ctx_ddl.add_stop_section('xmldatagroup', 'VAL');
     ctx_ddl.add_stop_section('xmldatagroup', 'COL1');
     ctx_ddl.add_stop_section('xmldatagroup', 'COL2');
end;
create index myindex on table1(col1)
indextype is ctxsys.context
parameters ('datastore ctxsys.default_datastore
filter ctxsys.null_filter
section group xmldatagroup')
But when I compared these indexes I found that sizes are the same and the token's table DR$MYINDEX$I for the second index contains tokens from tags from stop sections however in documentation I see:
Stop sections, empty tags, processing instructions, and comments are not indexed.
Is anybody know what's wrong with my code or stop section is just logical declaration for using on the query's execution step?

Answer 1

I'm not sure why your stop sections aren't working, but if you know the layout of the document I would recoomend using explcitly-defined field sections rather than auto_section_group.
This will be a lot more efficient for querying.
It will still index the data outside of the <NAME> section, though as part of the "default" section. If you don't want to index that, you could use a user_datastore to remove all the stuff apart from what you want to index.
- Roger