Although I haven't tried it myself, I did a search for Á´³ÑȾ³ÑÊÑ´¹ and
found this page.
It appears people use jcode and tr to solve this problem.

http://www.eml.ele.cst.nihon-u.ac.jp/~momma/wiki/wiki.cgi/Ruby/%E5%85%A8%E8%A7%92%E5%8D%8A%E8%A7%92%E5%A4%89%E6%8F%9B.html
http://blog.grayproductions.net/articles/the_kcode_variable_and_jcode_library


2009/6/25 Eric Hodel <drbrain / segment7.net>:
> On Jun 25, 2009, at 14:29, Ad Ad wrote:
>
>> Hi,
>> I am retrieving a string from a txt file.
>> The file contains some utf8 characters.
>>
>> I am comparing these characters against a default string.
>>
>> The problem is that some of the characters are not stored in a default
>> format.
>>
>> For example:
>> A is stored as £Á
>>
>> Naturally when I compare the character it fails.
>> Strangely when I unpacked the character it appears as 65313 which is the
>> correct utf8 number for A.
>>
>> Any way around this?
>
> Well, £Á is "Fullwidth Latin Capital Letter A" from the "Hiragana and
> Katakana" category (Unicode FF21) whereas A is "Latin Capital Letter A" from
> the "Latin" category (Unicode 0041).
>
> I don't know of a way to translate between the two categories, but maybe
> that will help.
>

-- 
Cheers,
 James Rubingh
http://www.wrive.com