Search theory ...

Question

Hi all
I'm not asking for source. I write code cos it's fun. But I do need some help with some thinks. This is what I need to do:
Essentially, I want to compare two strings and work out, to a percentage, how similar they are. The behaviour of my application thereafter is based on this percentage - will probably bug the user to retype their input if the % is too small. The idea is to allow a substantial margin for user error - to work out if what they typed was probably something I already have in a list somewhere else.
The question is, how to go about comparing for same-ness.
Initially I thought, convert both strings to the same case and strip the spaces. This makes a better basis for content comparison, as well as introducing a small level of idiot-proofing. This has worked well for me in the past.
Next, I thought maybe count the occurance of each letter - how many a's, how many b's, etc and make a comparison based on that, but that is actually no good, because it's possible to have totally different strings made of the same letters (duh).
So then I went back to scratch: convert to same case, but leave spaces as this allows me to do a word by word comparison. I could see if most of the word

Answer 1

Hi tommeke100
I like what you're saying about the hamming distance. I could definitely implement something like that, and it would work wonders for the word by word comparison.
It falls down though, when you consider that the strings will always consist of multiple words. An extra space or letter in the wrong place, if I understand what you've written correctly, will end up in a MASSIVE hamming distance. So to use the hamming effectively, I'd have to strip the spaces again, which I'm unwilling to do because the hamming distance could be misreported in the case of one spelling error somewhere.
So Hamming is only effective on individual words, but not only strings containing a whole sentence.
Perhaps what I could do is Both: First do the Hamming on the whole string. If I get a zero, it's obviously the same string, move on. Greater than zero, then do the boolean word-by-word compare as well.
Hmmm.
What do you think?
This all supposes that I've correctly understood your Hamming distance approach.
You say HD("azerty", "azarty") = 1.
What does HD("azerty", "ayzert") resolve to?
(obvious I've never heard of Hamming distance, eh?)
Am I totally off track? How does character order affect Hamming distance?
Thanks,
TheFr00n