README file for the Czech-English Dependency Treebank, version 1.0
5/19/2004

-------------------------------------------------------------------
Authors: Martin Cmejrek, Jan Curin, Jan Hajic, Jiri Havelka, 
	 Vladislav Kubon, Zdenek Zabokrtsky

Project Leader: Jan Hajic

Address:
 Center for Computational Linguistics and
 Institute of Formal and Applied Linguistics

 Malostranské nám. 25
 118 00 Praha 1
 Czech Republic

 http://ckl.mff.cuni.cz
 http://ufal.mff.cuni.cz
 phone: +420 221 914 278
 fax: +420 221 914 309
 e-mail: pcedt@ufal.mff.cuni.cz

License:
 http://ufal.mff.cuni.cz/pcedt/pcedt1.0-register.html

(c) 2002-2004 Center for Computational Linguistics 
    and Institute of Formal and Applied Linguistics
-------------------------------------------------------------------

INTRODUCTION

Prague Czech-English Dependency Treebank version 1.0 (PCEDT 1.0) is a
corpus of Czech-English parallel resources suitable for experiments in
machine translation, with a special emphasis on dependency-based
(structural) translation (with evaluation data provided for
Czech-to-English systems).

DOCUMENTATION

doc/PCEDT_main.html
 ... browsable documentation of the CD

doc/PCEDT_license.html
 ... PCEDT 1.0 license

doc/csts.html
 ... link to csts document type description

doc/fs.html
 ... FS-format description

doc/README.txt
 ... this file

DATA

Formats of data on PCEDT 1.0 are FS-format, CSTS-format (read
description of these formats in doc/ directory), or raw
text. Character encoding for Czech texts is ISO-8859-2 (English texts
are in ASCII).

Use TrEd or NetGraph to view trees in both, CSTS and FS formats.

Brief description for individual data packages follows, for more
detailed description please refer to separate README.txt files located
in appropriate directories.

Czech-English Parallel Penn Treebank Corpus
-------------------------------------------

For the purposes of comarable evaluation data in this corpus are
divided into three sets: development, evaluation and training


data/PTB_corpus/original/En_[development|evaluation|training]
 ... original PTB-style annotation of English
 
data/PTB_corpus/raw/[Cz|En]_[development|evaluation|training]
 ... 49k sentences of English raw texts and 21k sentences of its Czech
 translations

data/PTB_corpus/reference_translations/En_[development|evaluation]
 ... 4 reference retranslations from Czech to English of 500 sentences

data/PTB_corpus/NIST_format/[Cz|En]_[development|evaluation|training]
 ... 22k parallel sentences in NIST mt evaluation format (dtd/mt.dtd)

data/PTB_corpus/automatic_tagged/Cz_[development|evaluation|training]
 ... morphologically analyzed and automatically tagged Czech translations

data/PTB_corpus/automatic_AR/En_[development|evaluation|training]
 ... automatic conversion of English part into analytical representation

data/PTB_corpus/automatic_AR/Cz_[development|evaluation|training]
 ... automatically parsed Czech part, by Collins and Charniak parsers
 for Czech

data/PTB_corpus/automatic_TR/En_[development|evaluation|training]
 ... automatic conversion of English part into tectogrammatical representation

data/PTB_corpus/automatic_TR/Cz_[development|evaluation|training]
 ... automatic conversion of Czech AR trees into tectogrammatcal representation

data/PTB_corpus/manual_TR/En_[development|evaluation|training]
 ... manual annotation on tectogrammatical level of 1275 English sentences

data/PTB_corpus/manual_TR/En_[development|evaluation|training]
 ... manual annotation on tectogrammatical level of 515 Czech sentences


Other Text Corpora
------------------

data/RD_corpus/raw/[Cz|En|Align]
 ... Reader's Digest parallel corpus

data/Czech_raw_text/
 ... 39M-word Czech monolingual corpus (tokenized, in csts-format)


Translation Dictionaries
------------------------

data/Dictionaries/
 ... Czech-English Translation Dictionaries


TOOLS

For instructions how to install and use tools included on the CD
please refer to:

tools/TrEd/Doc/TrEd.html for TrEd
tools/NetGraph/Doc/netgraph_manual.html for NetGraph client
tools/NetGraph/Doc/netgraph_server_install.html for NetGraph server
tools/SMT_QuickRun/Doc/SMT_QuickRun.html for SMT Quick Run Package