From: Peter Szinek <peter@...>
Date: 2006-04-06T23:11:04+09:00
Subject: Re: String generalization

Hi,

> Peter, thanks for the quiz. I needed it right now :-)
Actually it was not ment as quiz (its a task at hand at my workplace), 
but i like the mentionality of the Ruby ML that everybody started to 
solve it as a quiz ;-) Good work James!
> 
> I'm not sure I understand the rules. I had expected:
> 
>   '//' -> /+
Yeah, you re right with this one.
>   'jd�;d:L348kddd3' -> [a-z]+�;[a-z]:[A-Z]\d+[a-z]+\d
The idea is, that if the input string is a mixture of this and that
without some 'sensible' pattern, it should be just generalized to .*
Well, it is not very straightforward to pin down whether a string is a 
mixture or not, and when there is a sensible pattern in it. (And what is 
sensible in this context) ATM i am using some crazy heuristics wich 
works very nicely in the practice, but no one knows (not even me) what 
it is really doing ;-) It just works.

To put it another way: a human user can tell that 'jd�;d:L348kddd3' is 
some crap and not a repeating pattern of something (like a tel #, an 
email, a name) in any way. So the rule is, that if you, as a human user 
think it is a crap rather than something meaningful, the computer should 
say this as well ;-). And you produced a chunk of a code which can be 
wired into the first machine passing the Turing test...

bw,
Peter