Basically speaking, to build the alpha search, you pass the name to the
SSA-NAME3 algorithm It returns 5 byte tokens (binary fields) which you
store (along with the key to the table where the name is stored) in a
separate alpha search table. Now when you want to search on a name, you
first pass it through the SSANAME3 algorithm, then search the token table
to find the name matches. The nice thing about the product is that it
returns both exact name matches and phonetic name matches. For instance, if
you didn't know how to spell my name, (Szymanski), and input it as
Shamanske, the alpha search would bring my name back. The product had a
definite positive impact in reducing our alpha search query performance.
And no, I don't work for them.
Greg Szymanski
Wausau Insurance
Glenn Mackey <mac...@IBM.NET> on 01/28/99 04:39:59 PM
Please respond to DB2 Data Base Discussion List <DB...@AMERICAN.EDU>
Sent by: Glenn Mackey <mac...@IBM.NET>
To: DB...@AMERICAN.EDU
cc: (bcc: Greg Szymanski/Wausau/NWIE)
Subject: Names and Indexes ...
Hi,
I have a application which wants to do search on last name, but the length
of the column will be in the order of 80 - 100 bytes.
I just want a reality check on what other sites have done.
Do sites create indexes that long. The table will not be excessive,
~300,000 rows getting up to 1,000,000 in time.
TIA
Glenn Mackey
515 267 5767
Steve McMahon
This post does not reflect the opinions of either my client or employer
> ----------
> From: gszy...@WAUSAU.COM[SMTP:gszy...@WAUSAU.COM]
> Sent: Friday, January 29, 1999 8:58 AM
> To: DB...@AMERICAN.EDU
> Subject: Re: Names and Indexes ...