> I want to count the number of such encoded url,and
> classify it to malicious and normal url by its
> content. Then i'd see if url encoding especially
> host encoding could be used for url classification.

RFC2616:
   Characters other than those in the "reserved" and "unsafe" sets (see
   RFC 2396 [42]) are equivalent to their ""%" HEX HEX" encoding.

   For example, the following three URIs are equivalent:

      <http://abc.com:80/~smith/home.html>
      <http://ABC.com/%7Esmith/home.html>
      <http://ABC.com:/%7esmith/home.html>

By definition: not malicious and normal.