Page 1 of 1
Intelligent search function (MySQL+PHP)
Posted: Wed Aug 05, 2009 4:57 pm
by SvanteH
Is there an proven solution for finding out words that are similar (related?) to the original search-term of a database?
Lets say I search for %deaht%; is there a way of getting the similar word (Death) to be listed? I don't think SOUNDEX will give reliable results specially since the search will be based on another language than English.
Svante
Re: Intelligent search function (MySQL+PHP)
Posted: Thu Aug 06, 2009 9:08 am
by jackpf
Hmm....I don't know about soundex - never used it myself.
However, you could do something similiar to google with the "Did you mean....." with the pspell library and functions.
Re: Intelligent search function (MySQL+PHP)
Posted: Thu Aug 06, 2009 9:27 am
by VladSun
We had a related discussion here:
viewtopic.php?f=2&t=102560
Re: Intelligent search function (MySQL+PHP)
Posted: Thu Aug 06, 2009 9:32 am
by superdezign
Also, I suspect that Google saves all words that the users type and, if a word is misspelled often, it automatically uses the presumed correct spelling in the search. Google also determines what word you meant to type by context as well, which I assume is also by looking at common mistakes in searches and the corrections that the users give in a subsequent search.
Basically, Google is complex.
/rant
Re: Intelligent search function (MySQL+PHP)
Posted: Thu Aug 06, 2009 9:38 am
by jackpf
I said
like google. To
be google is impossible. Their coders are like....genii

Re: Intelligent search function (MySQL+PHP)
Posted: Thu Aug 06, 2009 9:43 am
by superdezign
jackpf wrote:genii
Ironically, it took me a second to realize what you said.
/stupidity
Re: Intelligent search function (MySQL+PHP)
Posted: Thu Aug 06, 2009 5:01 pm
by josh
superdezign wrote:Also, I suspect that Google saves all words that the users type and, if a word is misspelled often, it automatically uses the presumed correct spelling in the search. Google also determines what word you meant to type by context as well, which I assume is also by looking at common mistakes in searches and the corrections that the users give in a subsequent search.
Basically, Google is complex.
/rant
Sounds like a Bayesian network