Creating interMedia substring index on a CLOB that contains XML

Question

I have a CLOB column that has an existing XML_SECTION_GROUP index on certain tags within the XML content of the CLOB. (see script below)
begin
ctx_ddl.create_section_group('XMLDOC','XML_SECTION_GROUP');
end;
begin
ctx_ddl.add_zone_section ('XMLDOC','title','title');
ctx_ddl.add_zone_section('XMLDOC','keywords','keywords');
ctx_ddl.add_zone_section('XMLDOC','author','author');
end;
create index xmldoc_idx on xml_documents(xmldoc)
indextype is ctxsys.context
parameters ('section group xmldoc') ;
I have a requirement to search on the 'author' zone section by the first letter only. I figure a substring index is the way to go but I'm not sure of the syntax to create a substring index. Especially when I have already declared a SECTION_GROUP preference on this column and I would need to create a WORDLIST preference as well I think. Can someone help me out here?
null

Answer 1

Here is the answer from Oracle Text Help:
"the main problem here is that you cannot apply that fancy substring
processing just to the author section -- it will apply to everything,
which will probably blow up the index size.
Anything you do will require reindexing the documents, so you cannot
really get around having to rebuild the index entirely.
Here are various ways to tackle the problem:
1. do nothing. Query just like: Z% WITHIN AUTHOR
PRO: don't have to rebuild index
CON: query is slow. some queries cannot be executed due to
wildcard maxterms limits.
2. create a wordlist preference with PREFIX_INDEX set to TRUE,
PREFIX_MIN_LENGTH set to 1 and PREFIX_MAX_LENGTH set to 1.
Query looks like: Z% WITHIN AUTHOR
PRO: moderately fast query
CON: must be on 8.1.7 or above
get a lot of junk from words from other sections
3. 2) as above, plus make AUTHOR, KEYWORDS, TITLE field sections
PRO: faster query than 2)
CON: field sections are less flexible w.r.t. nesting and
repeating occurrences
4. use a user_datastore or procedure_filter to transform the data
so that <AUTHOR>Steven King</AUTHOR> becomes
<AUTHORINIT>AIK</AUTHORINIT><AUTHOR>Steven King<AUTHOR>
Use field section for AUTHORINIT and query becomes:
AIK within AUTHORINIT (I used AIK instead of just K so that you
don't have to make I and A non-stopwords)
PRO: fastest query, smallest index
CON: most work
have to massage data so slows down indexing"