I’d like to make a search feature that searches based on “sounds like” match.
For instance, lets say I have a company list that looks like this (lets say we live in Bizzaro world too):
- Acme
- Already allusion cite LTD
- All ready illusion site INC
- Apart assent
- Assent sight
(Or something simmilar with names… George or Jeorge ? “Yah-way”, or “ye-hova” ?)
When someone searches for something that “sounds like” the soundex("site") == S230, they should see results for “Sight” also.
As most people who’ve used soudnex before already know, normal substring matches obviously don’t do this.
I’m trying to work out in my head how to make a WHERE clause that can match based on this, so instead of a typical WHERE company LIKE input, I’d like to run a soundex. Obviously if I run soundex on the whole company name, I won’t be able to do substring searching (for example, a user searching “ALL” will never match a soundex of “All ready”). Soundex split on each word might not be worthwhile either, so I’m not sure running all combinations of a soundex is a good idea… or even if that’s going to be computationally feasible in a database with more than 1000 records.
Basically the interaction I want to have is when (in an office or something) Tom says to Sally “That name was something like Rebekkah Schwartzkopff” and it can be searched phonetically for a fuzzy match.
Obviously we’re going to run into issues with non-English named companies because of soudnex, but I’m will to compromise on this one.
I’d like to do this without adding anything to the database, or a stored procedure.
Very interesting question. I did a little poking around and found this:
http://www.codeproject.com/KB/database/dmetaphone4.aspx
I haven’t tested it myself but it seems like it would be worth checking out.
It would require you to add something to the database, but I don’t see how you can implement the functionality you want with built in SQL Server functionality…