From: | Stuart Bishop <zen(at)shangri-la(dot)dropbear(dot)id(dot)au> |
---|---|
To: | pgsql-general(at)postgresql(dot)org |
Subject: | Re: phonetic and/or synonym search |
Date: | 2001-10-11 00:53:46 |
Message-ID: | 6D06C1EE-BDE2-11D5-9897-000393031882@shangri-la.dropbear.id.au |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-general |
On Sunday, October 7, 2001, at 01:47 AM, Frank Joerdens wrote:
> Does anyone know how to do a phonetic and/or synonym search (this would
> be for the German language mostly)? What's the approach in theory?
The first thing you need to do is track down an algorithm that converts
a word into a code representing how the word sounds. This is language
specific,
and all the ones I know of are English specific or surname specific.
These
might do the job well enough in German, but you would need to test.
Examples
of the algorithms you are after are soundex (an implementation exists in
the contrib
directory of your postgresql source) and NYSIIS (I've got a C
implementation for
PostgreSQL if there is interest). Once you have reduced a word to its
code, you
can test if two words sound alike by simply comparing them
( nysiis('katie') == nysiis('city').
Examples on how to use this might be simply to create the relevant
function index on
your table, if your value contains only one word:
create index idx_blah on people (nysiis(surname))
You can then just use "nysiis('smith') = nysiis(surname)" in your where
clause of your SQL.
If you have multiple words in your value, you need triggers to split the
phrases into
words and store them in another table. You then use this table to
perform your phonetic
searches.
--
Stuart Bishop <zen(at)shangri-la(dot)dropbear(dot)id(dot)au>
From | Date | Subject | |
---|---|---|---|
Next Message | Martijn van Oosterhout | 2001-10-11 01:05:01 | Re: Error Codes? |
Previous Message | H. Wade Minter | 2001-10-11 00:05:53 | Re: Where to count |