I has been doing some research and benchmarking today and these are my conclusion ( so far ), i did try with 3 difetents methods:

Chainsaw's
String::Aprox
String::Similarity

My choise at this moment is String::Similarity, why?
Well first have a look at this benchmark:

Benchmark: timing 1000000 iterations of String::Aprox, String::Similar
+ity, Chainsaw...
String::Aprox:  5 wallclock secs ( 3.70 usr +  0.02 sys =  3.72 CPU) @
+ 268889.49/s
(n=1000000)
String::Similarity:  7 wallclock secs ( 7.38 usr +  0.00 sys =  7.38 C
+PU) @ 135593
.22/s (n=1000000)
Chainsaw: 15 wallclock secs (14.16 usr +  0.02 sys = 14.17 CPU) @ 7056
+6.65/
s (n=1000000)
[download]

In the benchmark i just called functions which did the basic operation of each solution.
Even when String::Aprox is fastest it doesn't make all the matchs that String::Similariy does, Chainsaw aproach does well with some matches but it is the slowest and some matches didn't come up
I just wanted you to know :)
My answer to my very own question would be:

Use String::Similarity

But i will continue testing.....
___________________________________________
Optimus magister, bonus liber