========================================================================
The EGYPT toolkit, Version 1.0.
Developed by the Statistical Machine Translation team, WS'99, CLSP/JHU.
http://www.clsp.jhu.edu/ws99/projects/mt
========================================================================
RELEASE HITORY:

Dec 10, 1999:	limited beta release.
Feb 23, 2000:   Official release version 1.0.

CONTENTS:

The EGYPT directory is organized such that it has the following
directory trees:

1. The "tools" directory tree: each generic (not language specific) tool that
we developed is stored in its own subdirectory under the tools
directory. In this release, the following tools are included:

whittle/        a tool for preparing and splitting the corpora
		generating *.snt file formats (the format required by GIZA).
GIZA/		the statistical models training program (developed in C++).
cairo/		the word alignment visualiztion tool (developed in Java).
cairoize	scripts for generating alignments files in *.aln format (the
		format required by cairo) (developed in C and perl).
misc/		some useful scripts for collecting corpus statistics.

NOTE: The stack decoder that we used during the workshop is not distributed with
this release. It still needs some more work and hopefully will be
released in the near future (keep visiting the web page).

2. bin: this directory contains symbolic links to binary files of each
	one of the tools.

3. doc:	this directory contains symbolic links to the tools README
	files.

4. There is one directory tree per language. Each language directory has
the following subdirectories: 

	corpora/	to keep the corpora for that language. 
	tools/		to keep tools specific to that language
			(e.g. tokenizers)
	dict/		to keep dictionaries (if any)
	work/		we suggest that you use this subdirectory to
			keep intermediate files generated by the
			software tools.


Currently, the following language subdirectories are included:

	arabic/		The Quran (Islam's holy book) in Arabic with English 
			translation. It was aligned based on chapters and 
			verse numbers. Verses are typically equivalent
			to sentences. We have not done any experiments
			with this corpora.


	english/	that contains english tokenizer and some
			monolingual text that was downloaded from 
			the UN web site. It was used for our Timorese
			(Tetun)	experiements.

	french/		The Language Data Consortium (LDC) has a large,
			sentence-aligned corpora (http://www.ldc.upen.edu/).
			Also, we are currently working on aligning 
			a large, unencumbered hansard parallel corpora that we
			plan to release in the future. However, this may take a	
			while. If you can't wait, then you can download
			the corpora from http://

	tetun/		A small Tetun-English parallel corpora. It was
			sentence-aligned manually.