You can search the lucene and solr mailing lists for "denormalize"
but the general response is to try one of:
1. de-normalize the data while indexing
- advantage: one query
- disadvantage: data repetition
2. use 2 indices
- advantage: no need for repetition; this is necessary anyway if
each query execution requires two-parts
- disadvantage: more work for the app., more resources, etc.
3. if feasible, consider using Boolean query with filters
http://wiki.apache.org/lucene-java/LuceneFAQ#head-
f70612c6e4670e7fa2d5aeef4710effc522d85e0
4. if the requirement is better suited to a combination lucene-for-
text and RDBMS-for-query solution, then go that way
You can
Good luck!
-h
----------------------------------------------------------------------
Hira, N.R.
Cognocys, Inc.
Catch up on the news. http://www.cognocys.com/prospector/news.html
On 18-Sep-2008, at 10:27 AM, Dino Korah wrote:
Hi All,
Scenario: I have 100 documents in an index and if these documents
fall into
10 mutualy exclusive set; And within that set one of them is the main
document.
Now if I am to search on the index and group the result on 10 mutually
exclusive sets. And if I have to display the result with fields
displayed
from the main document, How do I do that with minimal overhead.
Realworld Scenario: take the example of Inbox of emails with
attachments. If
those attachments and the email itself are indexed with a common
mail-id
field which is common for the attachments and the mail-body from
the same
mail message; Search results would be better displayed as email's
subject
and body extract so that user can identify them easily.
Many thanks
Dino
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]