Development of a German-English Translator

1
Development of a German-English Translator

• Felix Zhang
• Period 5
• 2007-2008
• Thomas Jefferson High School for Science and
  Technology Computer Systems Research Lab

2
Summary of Quarter 2

• NP Chunking
• Lemmatization
• Dictionary Lookup
• Inflection
• Noun-verb agreement

3
Scope for this quarter

• Focus less on statistical methods
• Get rudimentary grammar system working
• Fix all the bugs Ive made since September

4
New and Modified Components

• More info stored in NP chunking
• Better noun-verb agreement
• Grammar
• Element Assignment
• Priority Number Assignment

5
Noun-verb agreement

• Simple method to eliminate more ambiguities
• def eliminateother(attribs, sub, closest)
• for x in attribs
• if x01 "nou" and x ! sub
• for y in x1
• if y0 "nom"
• 
  attribsattribs.index(x)1.remove(y)?
• return attribs

6
Noun phrase chunking

• Now used for English sentences
• Stores more info for later methods
• the man make the children
• NP Chunked English 'the', 'art', 'man',
  'nou', 'akk', 'mas', 'dat', 'pl',
  'make', 'ver', '3', 'pl', 'pres', 'the',
  'art', 'small', 'adj', 'child', 'nou',
  'nom', 'pl'

7
Element Assignment

• Based on linguistic information
• If case is nominative, chunk is subject
• If accusative, chunk is direct object
• 'the', 'art', 'man', 'nou', 'akk', 'mas',
  'dat', 'pl', 'dobj', 'make', 'ver', '3',
  'pl', 'pres', 'mverb', 'the', 'art',
  'small', 'adj', 'child', 'nou', 'nom',
  'pl', 'sub'

8
Priority Assignment

• Each sentence element is assigned priority number
• Based on position in English sentence
• Assignments
• sub 1
• mverb 2
• auxverb 3
• iobj 4
• dobj 5
• Sort by number for English grammar

9
Full run of program

• input den Mann machen die kleinen Kinder
• The small children make the man
• fzhang_at_ltsp1 /research python proj.py
• Part of speech tags 'den', 'art', 'Mann',
  'nou', 'machen', 'ver', 'die', 'art',
  'kleinen', 'adj', 'Kinder', 'nou'
• Morphological analysis 'Mann', 'nou',
  'akk', 'mas', 'dat', 'pl', 'machen',
  'ver', '1', 'pl', '3', 'pl', 'pres',
  'kleinen', 'adj', 'nom', 'pl', 'akk',
  'pl', 'Kinder', 'nou', 'nom', 'pl',
  'akk', 'pl'
• Disambiguated after noun-verb agreement
  'Mann', 'nou', 'akk', 'mas', 'dat',
  'pl', 'machen', 'ver', '3', 'pl',
  'pres', 'kleinen', 'adj', 'nom', 'pl',
  'akk', 'pl', 'Kinder', 'nou', 'nom',
  'pl'
• Lemmatized 'Mann', 'Mann', 'Man',
  'machen', 'machen', 'kleinen', 'klein',
  'Kinder', 'Kind'
• Root translated 'den', 'the', 'Mann',
  'man', 'machen', 'make', 'die', 'the',
  'kleinen', 'small', 'Kinder', 'child'
• NP Chunked English 'the', 'art', 'man',
  'nou', 'akk', 'mas', 'dat', 'pl',
  'make', 'ver', '3', 'pl', 'pres', 'the',
  'art', 'small', 'adj', 'child', 'nou',
  'nom', 'pl'
• Inflected (only works before chunking)
• 'the', 'the' 'man', 'akk', 'mas', 'man'
  'man', 'dat', 'pl', 'mans' 'make', '3',
  'pl', 'make' 'the', 'the' 'small', 'small'
  'child', 'nom', 'pl', 'childs'
• Assigned an element type
• 'the', 'art', 'man', 'nou', 'akk', 'mas',
  'dat', 'pl', 'dobj', 'make', 'ver', '3',
  'pl', 'pres', 'mverb', 'the', 'art',
  'small', 'adj', 'child', 'nou', 'nom',
  'pl', 'sub'
• Assigned priority
• '5', 'the', 'art', 'man', 'nou', 'akk',
  'mas', 'dat', 'pl', 'dobj', '2', 'make',
  'ver', '3', 'pl', 'pres', 'mverb', '1',
  'the', 'art', 'small', 'adj', 'child',
  'nou', 'nom', 'pl', 'sub'
• Rearranged to English structure
• '1', 'the', 'art', 'small', 'adj',
  'child', 'nou', 'nom', 'pl', 'sub', '2',
  'make', 'ver', '3', 'pl', 'pres', 'mverb',
  '5', 'the', 'art', 'man', 'nou', 'akk',
  'mas', 'dat', 'pl', 'dobj'

10
Problems

• Ambiguities (again)?
• One ambiguity can change the entire structure of
  the sentence
• I gave a horse the hat vs. I gave the hat a
  horse
• Attempt at all permutations possible
• User disambiguation

11
Problems

• Inflexible
• Grammar can only be rearranged in one specific
  way
• Subject Main verb Indirect Direct
  Auxiliary Verb
• Does not accommodate for prepositions,
  conjunctions, etc.

12
Future research

• Implement more statistical methods
• Morphological info
• Actual translation bilingual corpus
• Create better parse tree Dependency grammar