Overview
Machine Translation (MT) is one of the oldest and still
far from solved challenges undertaken by computer science.
The course will present an overview of the history, approaches,
progress and difficulties of MT.
The central topic of the course will be the statistical MT (SMT)
approach introduced in the early 90's at IBM.
In particular, the following topics will be covered:
- statistical framework of MT
- word alignment models
- phrase-based translation
- log-linear models
- training and search algorithms
- minimum error training algorithms
- advanced topics:
- spoken language translation
- syntax-based SMT
- efficiency issues
- performance evaluation
- overview of publicly available software.
Alternative MT approaches will be discusses, and current research
trends in the field will be presented.
The course is part of the activities of the Dottorato in Informatica (PhD in
Informatics) at the Università di Pisa.
Bibliography
- P. Brown, S. Della Pietra, V. Della Pietra, R. Mercer, The Mathematics of
Statistical Machine Translation: Parameter Estimation,
Computational Linguistics, 1993.
- F. Och, H. Ney, A Systematic
Comparison of Various Statistical Alignment Models, Computational
Linguistics, 2003.
- U. Germann, M. Jahr, K. Knight, D. Marcu, K. Yamada, Fast
decoding and optimal decoding for machine translation, Proc. ACL,
2001.
- C. Tillmann, H. Ney, Word
Reordering and a Dynamic Programming Beam Search Algorithm for
Statistical Machine Translation, Computational Linguistics, 2003.
- K. Papineni, S. Roukos, T. Ward, Wei-Jing Zhu. BLEU: a
Method for Automatic Evaluation of Machine Translation, Proc. of
ACL 2002.
- P. Koehn, F.J. Och, D. Marcu. Statistical
Phrase-Based Translation. Proc. of HLT/NAACL 2003, 127-133.
- R. Zens, H. Ney, Improvements
in Phrase-Based Statistical Machine Translation, Proc. HLT-NAACL,
2004.
- F. Och, H. Ney, The
Alignment Template Approach to Statistical Machine Translation,
Computational Linguistics, 2004.
- M. Federico, N. Bertoldi, A Word-to-Phrase
Statistical Translation Model, ACM TSLP, 2005.
- P. Koehn et al., Moses: Open
Source Toolkit for Statistical Machine Translation, Proc. ACL
Demo&Poster Sessions, 2007.
- F. Och and H. Ney, Discriminative
training and maximum entropy models for statistical machine translation,
Proc. of ACL, 2002.
- F. Och, Minimum
Error Rate Training in Statistical Machine Translation, Proc. of
ACL, 2003.
- N. Bertoldi, R. Zens, M. Federico, Speech
Translation by Confusion Network Decoding, Proc. ICASSP, 2007.
- F. Casacuberta, M. Federico, H. Ney, E. Vidal. Recent
Efforts in Spoken
Language Translation, IEEE Signal Processing Magazine, 2008.
- C. Callison-Burch and M. Osborne and P. Koehn, Re-evaluation
the Role of Bleu in Machine Translation Research, Proc. EACL 2006.
Exam
Students are requested to prepare a written report reviewing one of the
above journal papers, or equivalently two conference papers about the
same topic.
Papers have to be selected with the advice of the teacher. The report
should have the following structure:
introduction to the topic
(500 words), explaining
its importance, difficulty, and possible solutions;
description of the approach(es) in the paper(s) (750 words);
presentation of the experimental results
(750 words): tasks, data sets, evaluation methods, results; conclusions
and considerations by the authors (500 words);
discussion by the student (500 words),
pointing out possible weak and strong points,
directions for improving or enhancing the work, and connections with the course
material. Reports have to be
written in English, with one the following Latex or Word
style
files, and sent as pdf files to the teacher.