Been reading thru all the posts and KB's on dealing with searching
across multiple columns. Most examples indicate you know what you're
looking for in each column.
Here's my big problem:
A: I have 12 tables, each with multiple columns that need to be
searched.
B: User will enter a search condition in any or all the boxes below
1 - "All these words"
2 - "Exact phrase"
3 - "Any of these words"
Unlike the SP3 "fix", if they enter "house cat" in "all these words", I
want a match if "house" and "cat" are in different tables.
In a nutshell, I need to take the 3 inputs, and search across all the
tables, and get ranked results.
The only answer I see, which seems to be kludge, is to create another
table with one column that concats all the indexable columns for all
the tables.
Doing a monster union with multiple contains/free/? on each select
seems really ugly and would kill performance.
Am I missing the boat, or does anyone have a better solution?
All comments are REALLY appreciated, as I just pulled out the last hair
on my head!
Thanks.
My approach would be to consolidate all of your data into a single table. I
realize it doesn't help much.
Hilary Cotter
Looking for a SQL Server replication book?
http://www.nwsu.com/0974973602.html
Looking for a FAQ on Indexing Services/SQL FTS
http://www.indexserverfaq.com
"woody" <shellymatthys@.yahoo.com> wrote in message
news:1116568453.936941.176620@.g47g2000cwa.googlegr oups.com...
> Been reading thru all the posts and KB's on dealing with searching
> across multiple columns. Most examples indicate you know what you're
> looking for in each column.
> Here's my big problem:
> A: I have 12 tables, each with multiple columns that need to be
> searched.
> B: User will enter a search condition in any or all the boxes below
> 1 - "All these words"
> 2 - "Exact phrase"
> 3 - "Any of these words"
> Unlike the SP3 "fix", if they enter "house cat" in "all these words", I
> want a match if "house" and "cat" are in different tables.
> In a nutshell, I need to take the 3 inputs, and search across all the
> tables, and get ranked results.
> The only answer I see, which seems to be kludge, is to create another
> table with one column that concats all the indexable columns for all
> the tables.
> Doing a monster union with multiple contains/free/? on each select
> seems really ugly and would kill performance.
> Am I missing the boat, or does anyone have a better solution?
> All comments are REALLY appreciated, as I just pulled out the last hair
> on my head!
> Thanks.
>
没有评论:
发表评论