NYCPHP Meetup

[Sterling Hughes]

On Fri, 2003-04-11 at 14:31, Chris Shiflett wrote:
> --- Mark Armendariz <nyphp at enobrev.com> wrote:
> > They want to eventually have something they called "fingerprint"
> > searching.  The term alone meant nothing to me (beyond the obvious -
> > which google agreed with).
> >  
> > Basically it has something to do with using synonyms and misspellings
> > and all types of other fun algorithms to find what a user is looking
> > for.  Does anyone hear have any experience with such things, or maybe
> > know where to look?
> 
> I'm not sure about synonyms, but support for misspelled words can typically be
> achieved with the Levenshtein algorithm, and PHP supports that:
> 
> http://www.php.net/manual/en/function.levenshtein.php
> 

Levenshtein is pretty useless in this case.  It'll tell you what words are semantically
equivalent, sure, but it doesn't do spelling correction.  I suggest you take a look 
at using pspell for spell checking.  As for synonyms, you need a pretty large dictionary
for that - take a  look freshmeat.

-Sterling

-- 
"Nothing is particularly hard if you divide it into small jobs." 
    - Henry Ford