================
Phone 2 Katakana
Brian Butz
================

============
REQUIREMENTS
============

This project requires a few gems:
  * rake
  * active_record (2.3.5)
  * sishen-rtranslate (1.2)
  * rdebug

  install any with: sudo gem install
  (might need to add github as a gem source with:
  	gem sources -a http:https://gems.github.com )

=====
USAGE
=====

Everything is run via rake tasks that can be found in Rakefile.

To initialize the database you'll want to run the following:
	
	rake migrate 			(might take up to 10 or so minutes)
	rake train TIMES=10000 	  (again, another 10 or so minutes)
	rake guess								  (and you are in!)

=====
TASKS
=====

	* migrate : Will migrate the database and import any
				corpus data, which ends up taking a bit of time

				Takes VERSION= option if you want to set the version

	* train	: This will train the data by creating intersections
			  using random words from the katakana corpus

			  Will need to run this for anything interesting to
			  happen

			  Takes TIMES= that defaults to 1000 (initially you'll
			  want 5k to 10k)

	* console : Using rdebug and the fact that environment loads
			    everything, this is sort of like a Rails console
			    for manipulating models and viewing the data

	* experiment1 : This experiment used the google translate gem
					and the CMU corpus to generate a corpus of
					all katakana words based on that corpus

					This will run for a few hours (though could
					most likely be opitmized to hit google only
					once instead of 140k times)

					It is not neccessary to run this

	* guess : If this project has a "main" here it is
			  This lets you enter a word and get back it's
			  phonetic structure, the raw and phonetic katakana
			  if it exists, and the raw and phonetic katakana
			  that the program guesses