机器翻译---SMT2007-1.ppt

Introduction toStatistical Machine Translation Lecture of Fall Semester 2007 for The Principle of Machine Translation Zhao Tiejun Machine Intelligence and Translation Laboratory Outline Lecture 1 – Brief introduction to basic ideas in SMT Lecture 2 – Statistical translation models: IBM models Lecture 3 – Word alignment and EM algorithm Lecture 4 – Phrase-based SMT Lecture 5 – Syntax-based SMT Lecture 1 Brief introduction to basic ideas in SMT References Philipp Koehn, 2007, Empirical Methods in Natural Language Processing, Lecture 14, http://www.inf.ed.ac.uk/teaching/courses/ emnlp Franz J. Och, 2005, Statistical Machine Translation: Foundations and Recent Advances, Tutorial at MT Summit 2005, /MTS-2005-Och.pdf Chris Callison-Burch Philipp Koehn, 2005, Introduction to Statistical Machine Translation, lecture on 17th European Summer School in Logic, Language and Information(ESSLLI) Franz Josef Och and Hermann Ney, 2002, Discriminative Training and Maximum Entropy Models for Statistical Machine Translation, Proc. of ACL 2002, pp295~302 Outline of Lecture 1 1.1 Machine translation: different views 1.2 Parallel texts and SMT basic formula 1.3 Translation results and automatic evaluation 1.4 MT resources and community Appendix: story of Rosetta stone Machine translation: different views (1) What is machine translation? Academic definitions: the attempt to automate all or part of the process of translating from one human language to another (Arnold et al. 1993) apply computers to translate texts of one natural language into texts of another natural language (Hutchines, 1986) Processed objects: human language Formats: text, speech Tools: computers of every types Procedure: automatic or semi-automatic Machine translation: different views (2) From the view of a user, what does he expect: Given a source language token, any MT engine should:- 1. Output the token as an appropriate string or substring in the target language. 2. Output the source language token if unkn