Subject Re: [IBO] Soundex and SoundexMax...
Author Dany M
Geoff Worboys wrote:
>> Secondly the way Jason included SoundexMax in the IB_FTS
>> utility.
> Hmmm... I have never used FTS. A quick glance at the code
> looks to me as though the SoundExMax facility was put in there
> without understanding how it was meant to be used.
> This is probably understandable to the extent that I believe
> Jason preferred the metaphone algorithm, so probably never
> really studied what it was I had done with soundex. I had
> already done my app using soundex so never really looked at
> whether the metaphone stuff was better or worse.

Yes, well... I have greater problems here. All these inventions are for
the American language :( I had a scientist friend willing to write
something tuned to Swedish names (or rather the way swedes tend to spell
and pronounce even foreign names) but he has a lot of other stuff to
tend to so theres no results there yet. My sporadic tests indicate
soundex is better for my use. It will be a breeze to rewrite the FTS
components query engine (I hope).

Jason, I made some changes to scrap the auxiliary field. It does not
seem to be of much use most of the times and the code works without it
(in FB 1.5, 2.0 will need some work because of the alias stuff). It also
seems to be no problem to use another unique non-segmented index in
stead of the PK for <<KEY_COL>>, you just need to load those into the
drop-down boxes. Also, the utility shows something lacking in the way it
loads the drop-down boxes when you are adding indices - I haven't been
able to remedy that. Just FYI if you want to put it on your list. If
not, I'm fine as is. If I judge my changes good enough to publish I'll
mail them to you but it is still WIP.

Thanks a lot,