Preintroduction to Machine Translation

1
(Pre-)introduction to Machine Translation

• LN3007Translating by computers

2
Topics

• What is machine translation?
• Why MT matters?
• Did you use machine translation?
• For what purposes?
• What was the quality?
• Did you find it useful?

3
What is machine translation?

• Machine translation (MT) is a form of translation
  where a computer program analyses the text in one
  language - the "source text" - and then produces
  another, equivalent text in another language -
  the target text - without human intervention.

4
Machine translation

• Currently the state of machine translation is
  such that it involves some human intervention, as
  it requires a pre-editing and a post-editing
  phase. Note that in machine translation, the
  translator supports the machine and not the other
  way around.
• Nowadays most machine translation systems produce
  what is called a gisting translation - a rough
  translation that gives the "gist" of the source
  text

5
ALPAC report
Weaver Memo
6
Why MT matters?

• Social and political importance
• Commercial importance
• Scientifically posses challenges to many fields
• Philosophically MT is an interesting matter

7
Case study 1

• A commercial company which produces hardware
• Technical manuals have to be produced in several
  languages, but the persons who write them know
  only one language
• Legal documents, email, correspondence has to be
  also translated
• Which is your solution?

8
Case study 1

• The (technical) documents available in electronic
  format are translated using MT
• Post edit the output
• Add the specific terms to the dictionary
• Review the translation (usually done by someone
  else)

9
Why technical documents?

• Because the text has to be written in a style
  which is suitable for MT
• Keep sentences short
• Make sure sentences are grammatical
• Avoid complicated grammatical senses
• Do not use polysemous words
• More guidelines

10
Babelize
Translate (by computer) an English phrase back
and forth between different languages e.g. The
spirit is willing, but the flesh is weak.
11
Case study 2

• Email correspondence in a multinational
  corporation
• Needs to work with email clients (e.g. Outlook)
• Needs to integrate with applications such as MS
  Word, etc.
• Needs to be totally transparent to the user

12
Case study 2 How it should work

• Send the email to the server
• The server determines the language of the source
  email and the language of the recipient
• Automatically translates the message (specific
  dictionaries are used)
• The sender has the option to disambiguate the
  polysemous words

13
Case study 3

• Translate webpages written in languages that you
  dont know
• Example
• It is good to give you the gist of the page

14
Case study 4

• Speech-to-speech translation
• Posses additional challenges because
  understanding of speech is quite a difficult
  problem