Difference between revisions of "Machine Translation"
From Cohen Courses
Jump to navigationJump to searchPastStudents (talk | contribs) |
PastStudents (talk | contribs) |
||
| Line 18: | Line 18: | ||
== Evaluation Methods == | == Evaluation Methods == | ||
| − | Machine translation systems have been evaluated in a number of ways, including human judgement and automated methods. Some popular automated methods of evaluation include [[ | + | Machine translation systems have been evaluated in a number of ways, including human judgement and automated methods. Some popular automated methods of evaluation include [[BLEU]], [[NIST_metric|NIST]], and [[METEOR]]. |
== Example Systems == | == Example Systems == | ||
Latest revision as of 15:46, 30 November 2010
Contents
Summary
Machine Translation (or MT for short) is a problem in the field of computational linguistics which looks at translating some input in one natural language, in the form of text or speech, into another natural language with computer algorithms/software.
Common Approaches
Some common approaches to Machine Translation include the following:
- Rule-based, includes transfer-based, interlingual, and dictionary-based translations
- Statistical, generates translations by using statistical methods on bilingual corpora
- Example Based, basically "translation by analogy"
- Hybrid, combines aspects of both rule-based and statistical machine translation
Challenges / Issues
Some major challenges in machine translation include handling named entities, word sense disambiguation, and handling special cases like idioms and out of vocabulary words.
Evaluation Methods
Machine translation systems have been evaluated in a number of ways, including human judgement and automated methods. Some popular automated methods of evaluation include BLEU, NIST, and METEOR.
Example Systems
References / Links
- Wikipedia article on Machine Translation - [1]
