Can I process UTF-8 datasets/files?

Java can process UTF-8 files without any problems, it is just that Java uses a different encoding for displaying them under Windows (= "Cp1252"). If you change the file encoding to "utf-8" you'll be fine (discussed in this WEKAlist post).

If you are running WEKA directly from the commandline, just add the following parameter to your commandline:
  • -Dfile.encoding=utf-8
If you are starting WEKA from the Start menu, then edit the RunWEKA.ini file:
  • If a fileEncoding placeholder already exists, then just change the value from "Cp1252" to "utf-8" (without the quotes of course).
  • If there isn't a fileEncoding yet, just the -Dfile.encoding=utf-8 parameter to all the java/javaw commands (see example RunWEKA.ini in this post).