That seems to be quite interesting - however as you pointed out, may not quite work with proper nouns, in particular from many languages ... Thanks anyway!
2009/4/4 Mark Rickerby <[email protected]> > > Here's a suggestion: > > While not quite your exact use case (you're dealing with proper nouns > - ie: company names) the common term for algorithms that convert > different forms of a word into its root term is called "stemming". The > Porter stemming algorithm is the most widely known and there are > various libraries available. > > Spelling correction requires more of a statistical/probability based > approach. It involves 'training' the corrector to recognize correct > words and rearranging the construction of an incorrect word to find > what the closest correct spelling might be. The classic article by > Peter Norvig of Google is a good start: > > http://norvig.com/spell-correct.html > > > > --~--~---------~--~----~------------~-------~--~----~ NZ PHP Users Group: http://groups.google.com/group/nzphpug To post, send email to [email protected] To unsubscribe, send email to [email protected] -~----------~----~----~----~------~----~------~--~---
