I have already proposed this solution, except it would require a
modification of the db to feed the data, a modification to the indexing
process, and modification of the searching.
That didn't go well with management, so I'm wondering if there's a way
to do this with only having to modify the searcher. :)
From: Erik Hatcher
Sent: Thursday, December 07, 2006 11:35 AM
Subject: Re: Search and Uniqueness
Why not just allow documents to have multiple category values rather
than duplicating documents?
One document per itemid seems all you'd need to me.
On Dec 7, 2006, at 11:09 AM, Filip Stanek wrote:
We have an index consisting of about 10 million items, so iteration
through records is not possible.
The index has a "description" field that is searched, it also contains
an "itemid" field, and it contains a "categorytree" field.
Each item in the index may belong to more than one category, so each
item may appear in the index more than once. However, when the
searched, we do not want to pull duplicate items.
Would there be a way to search on the "description" field but only
return items which have unique "itemid"?