C SC 620 Advanced Topics in Natural Language Processing

1
C SC 620Advanced Topics in Natural Language
Processing

• Sandiway Fong

2
Student Presentations Today

• Ben Wing Cross-Linguistic Discovery of Semantic
  Regularity. Peters, W. et al.

3
Machine Translation

• Readings in Machine Translation, Eds. Nirenburg,
  S. et al. MIT Press 2003.
• Part 1 Historical Perspective
• Reading list
• Introduction. Nirenburg, S.
• 1. Translation. Weaver, W.
• 3. The Mechanical Determination of Meaning.
  Reifer, E.
• 5. A Framework for Syntactic Translation. Yngve,
  V.
• 6. The Present Status of Automatic Translation of
  Languages. Bar-Hillel, Y.

4
Paper 1 Translation, W. Weaver

• A War Anecdote - Language Invariants
• I was amazed to discover that, for (apparently)
  widely varying languages, the basic logical
  structures have important common features -
  citing Reichenbach
• Modern comment
• In linguistic theory, we have the Universal
  Grammar (UG) hypothesis. The idea there is a
  dedicated language faculty of the human brain.
  Core language principles and constraints are the
  same, languages exhibit parametric differences
  with respect to word order, sentence
  constructions, etc.

5
Paper 1 Translation, W. Weaver

• Translation and Computers
• Weaver
• Semantic difficulties - multiple meanings
• Domain restrictions - scientific literature (less
  semantic difficulty).
• Rough translation - inelegant (but
  intelligible) still worthwhile
• Wiener
• Basic English is the reverse of mechanical and
  throws upon such words as get a burden which is
  much greater than most words carry
• Weaver
• Multiple meanings on get yes but a limited number
  of two word combinations get up, get over, get
  back
• 2000 words gt 4 million two word combinations -
  not formidable to a modern (1947) computer

6

• (make children "Abraham begot Isaac" "Men often
  father children but don't recognize them")
• (undergo (as of injuries and illnesses) "She
  suffered a fracture in the accident" "He had an
  insulin shock after eating three candy bars"
  "She got a bruise on her leg" "He got his arm
  broken in the scuffle")
• (be stricken by an illness, fall victim to an
  illness "He got AIDS" "She came down with
  pneumonia" "She took a chill")
• (come to have or undergo a change of (physical
  features and attributes) "He grew a beard" "The
  patient developed abdominal pains" "I got funny
  spots all over my body" "Well-developed
  breasts")
• (give certain properties to something "get
  someone mad" "She made us look silly" "He made
  a fool of himself at the meeting" "Don't make
  this into a big deal" "This invention will make
  you a millionaire" "Make yourself clear")
• (cause to move cause to be in a certain position
  or condition "He got his squad on the ball"
  "This let me in for a big surprise" "He got a
  girl into trouble")
• (enter or assume a certain state or condition
  "He became annoyed when he heard the bad news"
  "It must be getting more serious" "her face went
  red with anger" "She went into ecstasy" "Get
  going!")

7
Get in WN (as a verb) has 37 senses

• (take the first step or steps in carrying out an
  action "We began working at dawn" "Who will
  start?" "Get working as soon as the sun rises!"
  "The first tourists began to arrive in Cambodia"
  "He began early in the day")
• (receive a specified treatment (abstract) "These
  aspects of civilization do not find expression or
  receive an interpretation" "His movie received a
  good review" "I got nothing but trouble for my
  good intentions")
• (grasp with the mind "did you catch that
  allusion?" "We caught something of his theory in
  the lecture" "don't catch your meaning" "did
  you get it?" "She didn't get the joke")
• (be a mystery or bewildering to "This beats
  me!" "Got me--I don't know the answer!" "a
  vexing problem")
• (reach by calculation "What do you get when you
  add up these numbers?")
• (cause to do cause to act in a specified manner
  "The ads induced me to buy a VCR" "My children
  finally got me to buy a computer" "My wife made
  me buy a new sofa")

8
Paper 1 Translation, W. Weaver

• Translation and Computers
• A translation procedure that does little more
  than handle a one-to-one correspondence of words
  cannot hope to be useful for problems of literary
  translation 23rd Psalm vs. technical material,
  in which style is important, and in which the
  problems of idiom, multiple meanings, etc., are
  frequent.
• The Future of Computer Translation
• Look at four topics
• Meaning and Context
• What is the window size N to determine whether
  fast means rapid or motionless?
• Statistical semantic character of language
• Microcontext to settle difficult cases of
  ambiguity

9
Paper 1 Translation, W. Weaver

• The Future of Computer Translation
• Language and Logic
• Theorem McCulloch Pitts, 1943 A computer
  contructed with regenerative loops of a certain
  formal character is capable of deducing any
  legitimate conclusion from a finite set of
  premises
• Implies
• as written language is an expression of
  logical character, this theorem assures one that
  the problem is at least formally solvable
• Translation and Cryptography
• Citing Shannons work, he asks If we have
  useful methods for solving almost any
  cryptographic problem, may it not be that with
  proper interpretation we already have useful
  methods for translation?

10
Paper 1 Translation, W. Weaver

• The Future of Computer Translation
• Language and Invariants
• Most promising approach
• Thus may it be true that the way to translate
  from Chinese to Arabic, or from Russian to
  Portuguese, is not to attempt the direct route,
  shouting from tower to tower. Perhaps the way is
  to descend, from each language, down to the
  common base of human communication - the real but
  as yet undiscovered universal language - and then
  re-emerge by whatever particular route is
  convenient.
• Modern comment
• Interlingua or proto-language?