AVENUELETRAS: Learningbased MT Approaches for Languages with Limited Resources

1
AVENUE/LETRASLearning-based MT Approaches for
Languages with Limited Resources

• Alon Lavie
• Language Technologies Institute
• Carnegie Mellon University
• Joint work with
• Jaime Carbonell, Lori Levin, Kathrin Probst, Erik
  Peterson, Christian Monson, Ariadna Font-Llitjos,
  Alison Alvarez, Roberto Aranovich

2
Outline

• Rationale
• Framework overview
• Elicitation
• Learning transfer Rules
• Automatic rule refinement
• Example prototypes
• Implications for MT with vast parallel data
• Conclusions and future directions

3
Why Machine Translation for Languages with
Limited Resources?

• We are in the age of information explosion
• The internetwebGoogle ? anyone can get the
  information they want anytime
• But what about the text in all those other
  languages?
• How do they read all this English stuff?
• How do we read all the stuff that they put
  online?
• MT for these languages would Enable
• Better government access to native indigenous and
  minority communities
• Better minority and native community
  participation in information-rich activities
  (health care, education, government) without
  giving up their languages.
• Civilian and military applications (disaster
  relief)
• Language preservation

4
CMUs AVENUE Approach

• Elicitation use bilingual native informants to
  create a small high-quality word-aligned
  bilingual corpus of translated phrases and
  sentences
• Building Elicitation corpora from feature
  structures
• Feature Detection and Navigation
• Transfer-rule Learning apply ML-based methods to
  automatically acquire syntactic transfer rules
  for translation between the two languages
• Learn from major language to minor language
• Translate from minor language to major language
• XFER Decoder
• XFER engine produces a lattice of possible
  transferred structures at all levels
• Decoder searches and selects the best scoring
  combination
• Rule Refinement refine the acquired rules via a
  process of interaction with bilingual informants
• Morphology Learning
• Word and Phrase bilingual lexicon acquisition

5
AVENUE Architecture
6
The Transfer Engine
7
The Transfer Engine

• Some Unique Features
• Works with either learned or manually-developed
  transfer grammars
• Handles rules with or without unification
  constraints
• Supports interfacing with servers for
  Morphological analysis and generation
• Can handle ambiguous source-word analyses and/or
  SL segmentations represented in the form of
  lattice structures

8
The Lattice Decoder

• Simple Stack Decoder, similar in principle to
  SMT/EBMT decoders
• Searches for best-scoring path of non-overlapping
  lattice arcs
• Scoring based on log-linear combination of
  scoring components (no MER training yet)
• Scoring components
• Standard trigram LM
• Fragmentation how many arcs to cover the entire
  translation?
• Length Penalty
• Rule Scores (not fully integrated yet)

9
Outline

• Rationale for learning-based MT
• Roadmap for learning-based MT
• Framework overview
• Elicitation
• Learning transfer Rules
• Automatic rule refinement
• Example prototypes
• Implications for MT with vast parallel data
• Conclusions and future directions

10
Data Elicitation for Languages with Limited
Resources

• Rationale
• Large volumes of parallel text not available ?
  create a small maximally-diverse parallel corpus
  that directly supports the learning task
• Bilingual native informant(s) can translate and
  align a small pre-designed elicitation corpus,
  using elicitation tool
• Elicitation corpus designed to be typologically
  and structurally comprehensive and compositional
• Transfer-rule engine and new learning approach
  support acquisition of generalized transfer-rules
  from the data

11
Elicitation Tool English-Chinese Example
12
Elicitation ToolEnglish-Chinese Example
13
Elicitation ToolEnglish-Hindi Example
14
Elicitation ToolEnglish-Arabic Example
15
Elicitation ToolSpanish-Mapudungun Example
16
Designing Elicitation Corpora

• What do we want to elicit?
• Diversity of linguistic phenomena and
  constructions
• Syntactic structural diversity
• How do we construct an elicitation corpus?
• Typological Elicitation Corpus based on
  elicitation and documentation work of field
  linguists (e.g. Comrie 1977, Bouquiaux 1992)
  initial corpus size 1000 examples
• Structural Elicitation Corpus based on
  representative sample of English phrase
  structures 120 examples
• Organized compositionally elicit simple
  structures first, then use them as building
  blocks
• Goal minimize size, maximize linguistic coverage
  

17
Typological Elicitation Corpus

• Feature Detection
• Discover what features exist in the language and
  where/how they are marked
• Example does the language mark gender of nouns?
  How and where are these marked?
• Method compare translations of minimal pairs
  sentences that differ in only ONE feature
• Elicit translations/alignments for detected
  features and their combinations
• Dynamic corpus navigation based on feature
  detection no need to elicit for combinations
  involving non-existent features

18
Typological Elicitation Corpus

• Initial typological corpus of about 1000
  sentences was manually constructed
• New construction methodology for building an
  elicitation corpus using
• A feature specification lists inventory of
  available features and their values
• A definition of the set of desired feature
  structures
• Schemas define sets of desired combinations of
  features and values
• Multiplier algorithm generates the comprehensive
  set of feature structures
• A generation grammar and lexicon NLG generator
  generates NL sentences from the feature structures

19
Structural Elicitation Corpus

• Goal create a compact diverse sample corpus of
  syntactic phrase structures in English in order
  to elicit how these map into the elicited
  language
• Methodology
• Extracted all CFG rules from Brown section of
  Penn TreeBank (122K sentences)
• Simplified POS tag set
• Constructed frequency histogram of extracted
  rules
• Pulled out simplest phrases for most frequent
  rules for NPs, PPs, ADJPs, ADVPs, SBARs and
  Sentences
• Some manual inspection and refinement
• Resulting corpus of about 120 phrases/sentences
  representing common structures
• See Probst and Lavie, 2004

20
Outline

• Rationale for learning-based MT
• Roadmap for learning-based MT
• Framework overview
• Elicitation
• Learning transfer Rules
• Automatic rule refinement
• Example prototypes
• Implications for MT with vast parallel data
• Conclusions and future directions

21
Transfer Rule Formalism
SL the old man, TL ha-ish ha-zaqen NPNP
DET ADJ N -gt DET N DET ADJ ( (X1Y1) (X1Y3)
(X2Y4) (X3Y2) ((X1 AGR) 3-SING) ((X1 DEF
DEF) ((X3 AGR) 3-SING) ((X3 COUNT)
) ((Y1 DEF) DEF) ((Y3 DEF) DEF) ((Y2 AGR)
3-SING) ((Y2 GENDER) (Y4 GENDER)) )

• Type information
• Part-of-speech/constituent information
• Alignments
• x-side constraints
• y-side constraints
• xy-constraints,
• e.g. ((Y1 AGR) (X1 AGR))

22
Transfer Rule Formalism (II)
SL the old man, TL ha-ish ha-zaqen NPNP
DET ADJ N -gt DET N DET ADJ ( (X1Y1) (X1Y3)
(X2Y4) (X3Y2) ((X1 AGR) 3-SING) ((X1 DEF
DEF) ((X3 AGR) 3-SING) ((X3 COUNT)
) ((Y1 DEF) DEF) ((Y3 DEF) DEF) ((Y2 AGR)
3-SING) ((Y2 GENDER) (Y4 GENDER)) )

• Value constraints
• Agreement constraints

23
Rule Learning - Overview

• Goal Acquire Syntactic Transfer Rules
• Use available knowledge from the source side
  (grammatical structure)
• Three steps
• Flat Seed Generation first guesses at transfer
  rules flat syntactic structure
• Compositionality Learning use previously learned
  rules to learn hierarchical structure
• Constraint Learning refine rules by learning
  appropriate feature constraints

24
Flat Seed Rule Generation
25
Flat Seed Rule Generation

• Create a flat transfer rule specific to the
  sentence pair, partially abstracted to POS
• Words that are aligned word-to-word and have the
  same POS in both languages are generalized to
  their POS
• Words that have complex alignments (or not the
  same POS) remain lexicalized
• One seed rule for each translation example
• No feature constraints associated with seed rules
  (but mark the example(s) from which it was
  learned)

26
Compositionality Learning
27
Compositionality Learning

• Detection traverse the c-structure of the
  English sentence, add compositional structure for
  translatable chunks
• Generalization adjust constituent sequences and
  alignments
• Two implemented variants
• Safe Compositionality there exists a transfer
  rule that correctly translates the
  sub-constituent
• Maximal Compositionality Generalize the rule if
  supported by the alignments, even in the absence
  of an existing transfer rule for the
  sub-constituent

28
Constraint Learning
29
Constraint Learning

• Goal add appropriate feature constraints to the
  acquired rules
• Methodology
• Preserve general structural transfer
• Learn specific feature constraints from example
  set
• Seed rules are grouped into clusters of similar
  transfer structure (type, constituent sequences,
  alignments)
• Each cluster forms a version space a partially
  ordered hypothesis space with a specific and a
  general boundary
• The seed rules in a group form the specific
  boundary of a version space
• The general boundary is the (implicit) transfer
  rule with the same type, constituent sequences,
  and alignments, but no feature constraints

30
Constraint Learning Generalization

• The partial order of the version space
• Definition A transfer rule tr1 is strictly more
  general than another transfer rule tr2 if all
  f-structures that are satisfied by tr2 are also
  satisfied by tr1.
• Generalize rules by merging them
• Deletion of constraint
• Raising two value constraints to an agreement
  constraint, e.g.
• ((x1 num) pl), ((x3 num) pl) ?
• ((x1 num) (x3 num))

31
Automated Rule Refinement

• Bilingual informants can identify translation
  errors and pinpoint the errors
• A sophisticated trace of the translation path can
  identify likely sources for the error and do
  Blame Assignment
• Rule Refinement operators can be developed to
  modify the underlying translation grammar (and
  lexicon) based on characteristics of the error
  source
• Add or delete feature constraints from a rule
• Bifurcate a rule into two rules (general and
  specific)
• Add or correct lexical entries
• See Font-Llitjos, Carbonell Lavie, 2005

32
Outline

• Rationale for learning-based MT
• Roadmap for learning-based MT
• Framework overview
• Elicitation
• Learning transfer Rules
• Automatic rule refinement
• Example prototypes
• Implications for MT with vast parallel data
• Conclusions and future directions

33
AVENUE Prototypes

• General XFER framework under development for past
  three years
• Prototype systems so far
• German-to-English, Dutch-to-English
• Chinese-to-English
• Hindi-to-English
• Hebrew-to-English
• Portuguese-to-English
• In progress or planned
• Mapudungun-to-Spanish
• Quechua-to-Spanish
• Arabic-to-English
• Native-Brazilian languages to Brazilian Portuguese

34
Challenges for Hebrew MT

• Paucity in existing language resources for Hebrew
• No publicly available broad coverage
  morphological analyzer
• No publicly available bilingual lexicons or
  dictionaries
• No POS-tagged corpus or parse tree-bank corpus
  for Hebrew
• No large Hebrew/English parallel corpus
• Scenario well suited for CMU transfer-based MT
  framework for languages with limited resources

35
Hebrew-to-English MT Prototype

• Initial prototype developed within a two month
  intensive effort
• Accomplished
• Adapted available morphological analyzer
• Constructed a preliminary translation lexicon
• Translated and aligned Elicitation Corpus
• Learned XFER rules
• Developed (small) manual XFER grammar as a point
  of comparison
• System debugging and development
• Evaluated performance on unseen test data using
  automatic evaluation metrics

36
(No Transcript)
37
Morphology Example

• Input word BWRH
• 0 1 2 3 4
• --------BWRH--------
• -----B-----WR--H--
• --B---H----WRH---

38
Morphology Example

• Y0 ((SPANSTART 0) Y1 ((SPANSTART 0)
  Y2 ((SPANSTART 1)
• (SPANEND 4) (SPANEND
  2) (SPANEND 3)
• (LEX BWRH) (LEX B)
  (LEX WR)
• (POS N) (POS
  PREP)) (POS N)
• (GEN F)
  (GEN M)
• (NUM S)
  (NUM S)
• (STATUS ABSOLUTE))
  (STATUS ABSOLUTE))
• Y3 ((SPANSTART 3) Y4 ((SPANSTART 0)
  Y5 ((SPANSTART 1)
• (SPANEND 4) (SPANEND
  1) (SPANEND 2)
• (LEX LH) (LEX
  B) (LEX H)
• (POS POSS)) (POS
  PREP)) (POS DET))
• Y6 ((SPANSTART 2) Y7 ((SPANSTART 0)
• (SPANEND 4) (SPANEND
  4)
• (LEX WRH) (LEX
  BWRH)
• (POS N) (POS
  LEX))
• (GEN F)
• (NUM S)

39
Sample Output (dev-data)

• maxwell anurpung comes from ghana for israel four
  years ago and since worked in cleaning in hotels
  in eilat
• a few weeks ago announced if management club
  hotel that for him to leave israel according to
  the government instructions and immigration
  police
• in a letter in broken english which spread among
  the foreign workers thanks to them hotel for
  their hard work and announced that will purchase
  for hm flight tickets for their countries from
  their money

40
Evaluation Results

• Test set of 62 sentences from Haaretz newspaper,
  2 reference translations

41
Hebrew-English Test Suite Evaluation
42
Outline

• Rationale for learning-based MT
• Roadmap for learning-based MT
• Framework overview
• Elicitation
• Learning transfer Rules
• Automatic rule refinement
• Learning Morphology
• Example prototypes
• Implications for MT with vast parallel data
• Conclusions and future directions

43
Implications for MT with Vast Amounts of Parallel
Data

• Phrase-to-phrase MT ill suited for long-range
  reorderings ? ungrammatical output
• Recent work on hierarchical Stat-MT Chiang,
  2005 and parsing-based MT Melamed et al, 2005
  Knight et al
• Learning general tree-to-tree syntactic mappings
  is equally problematic
• Meaning is a hybrid of complex, non-compositional
  phrases embedded within a syntactic structure
• Some constituents can be translated in isolation,
  others require contextual mappings

44
Implications for MT with Vast Amounts of Parallel
Data

• Our approach for learning transfer rules is
  applicable to the large data scenario, subject to
  solutions for several large challenges
• No elicitation corpus ? break-down parallel
  sentences into reasonable learning examples
• Working with less reliable automatic word
  alignments rather than manual alignments
• Effective use of reliable parse structures for
  ONE language (i.e. English) and automatic word
  alignments in order to decompose the translation
  of a sentence into several compositional rules.
• Effective scoring of resulting very large
  transfer grammars, and scaled up transfer
  decoding

45
Implications for MT with Vast Amounts of Parallel
Data

• Example
• ? ?? ? ??? ?? ? ??
• He freq with J Zemin Pres via
  phone
• He freq talked with President J Zemin over
  the phone

46
Implications for MT with Vast Amounts of Parallel
Data

• Example
• ? ?? ? ??? ?? ? ??
• He freq with J Zemin Pres via
  phone
• He freq talked with President J Zemin over
  the phone

NP1
NP2
NP3
NP1
NP2
NP3
47
Conclusions

• There is hope yet for wide-spread MT between many
  of the worlds language pairs
• MT offers a fertile yet extremely challenging
  ground for learning-based approaches that
  leverage from diverse sources of information
• Syntactic structure of one or both languages
• Word-to-word correspondences
• Decomposable units of translation
• Statistical Language Models
• AVENUEs XFER approach provides a feasible
  solution to MT for languages with limited
  resources
• Promising approach for addressing the fundamental
  weaknesses in current corpus-based MT for
  languages with vast resources

48
Future Research Directions

• Automatic Transfer Rule Learning
• Learning mappings for non-compositional
  structures
• Effective models for rule scoring for
• Decoding using scores at runtime
• Pruning the large collections of learned rules
• Learning Unification Constraints
• In the large-data scenario from large volumes
  of uncontrolled parallel text automatically
  word-aligned
• In the absence of morphology or POS annotated
  lexica
• Integrated Xfer Engine and Decoder
• Improved models for scoring tree-to-tree
  mappings, integration with LM and other knowledge
  sources in the course of the search

49
Future Research Directions

• Automatic Rule Refinement
• Morphology Learning
• Feature Detection and Corpus Navigation

50
(No Transcript)
51
Mapudungun-to-Spanish Example
English I didnt see Maria
Mapudungun pelafiñ Maria
Spanish No vi a María
52
Mapudungun-to-Spanish Example
English I didnt see Maria
Mapudungun pelafiñ Maria pe -la -fi -ñ Maria see
-neg -3.obj -1.subj.indicative Maria
Spanish No vi a María No vi a María neg see.1.sub
j.past.indicative acc Maria
53
pe-la-fi-ñ Maria
V
pe
54
pe-la-fi-ñ Maria
V
pe
VSuff
Negation
la
55
pe-la-fi-ñ Maria
V
pe
VSuffG
Pass all features up
VSuff
la
56
pe-la-fi-ñ Maria
V
pe
VSuffG
VSuff
object person 3
fi
VSuff
la
57
pe-la-fi-ñ Maria
V
VSuffG
pe
Pass all features up from both children
VSuffG
VSuff
fi
VSuff
la
58
pe-la-fi-ñ Maria
V
VSuffG
VSuff
pe
person 1 number sg mood ind
VSuffG
VSuff
ñ
fi
VSuff
la
59
pe-la-fi-ñ Maria
V
VSuffG
VSuffG
VSuff
pe
Pass all features up from both children
VSuffG
VSuff
ñ
fi
VSuff
la
60
pe-la-fi-ñ Maria
Pass all features up from both children
V
Check that 1) negation 2) tense is undefined
V
VSuffG
VSuffG
VSuff
pe
VSuffG
VSuff
ñ
fi
VSuff
la
61
pe-la-fi-ñ Maria
NP
V
VSuffG
person 3 number sg human
VSuffG
VSuff
N
pe
VSuffG
VSuff
Maria
ñ
fi
VSuff
la
62
pe-la-fi-ñ Maria
S
Check that NP is human
Pass features up from
VP
NP
V
VSuffG
VSuffG
VSuff
N
pe
VSuffG
VSuff
ñ
Maria
fi
VSuff
la
63
Transfer to Spanish Top-Down
S
S
VP
VP
NP
V
VSuffG
VSuffG
VSuff
N
pe
VSuffG
VSuff
ñ
Maria
fi
VSuff
la
64
Transfer to Spanish Top-Down
Pass all features to Spanish side
S
S
VP
VP
NP
NP
a
V
VSuffG
VSuffG
VSuff
N
pe
VSuffG
VSuff
ñ
Maria
fi
VSuff
la
65
Transfer to Spanish Top-Down
S
S
Pass all features down
VP
VP
NP
NP
a
V
VSuffG
VSuffG
VSuff
N
pe
VSuffG
VSuff
ñ
Maria
fi
VSuff
la
66
Transfer to Spanish Top-Down
S
S
Pass object features down
VP
VP
NP
NP
a
V
VSuffG
VSuffG
VSuff
N
pe
VSuffG
VSuff
ñ
Maria
fi
VSuff
la
67
Transfer to Spanish Top-Down
S
S
VP
VP
NP
NP
a
V
VSuffG
Accusative marker on objects is introduced
because human
VSuffG
VSuff
N
pe
VSuffG
VSuff
ñ
Maria
fi
VSuff
la
68
Transfer to Spanish Top-Down
S
S
VP
VP
VPVP VBar NP -gt VBar "a" NP ( (X1Y1) (X2
Y3) ((X2 type) (NOT personal)) ((X2
human) c ) (X0 X1) ((X0 object) X2)
(Y0 X0) ((Y0 object) (X0 object)) (Y1
Y0) (Y3 (Y0 object)) ((Y1 objmarker person)
(Y3 person)) ((Y1 objmarker number) (Y3
number)) ((Y1 objmarker gender) (Y3 ender)))
NP
NP
a
V
VSuffG
VSuffG
VSuff
N
pe
VSuffG
VSuff
ñ
Maria
fi
VSuff
la
69
Transfer to Spanish Top-Down
S
S
Pass person, number, and mood features to Spanish
Verb
VP
VP
NP
NP
a
Assign tense past
V
VSuffG
V
no
VSuffG
VSuff
N
pe
VSuffG
VSuff
ñ
Maria
fi
VSuff
la
70
Transfer to Spanish Top-Down
S
S
VP
VP
NP
NP
a
V
VSuffG
V
no
VSuffG
VSuff
N
pe
VSuffG
VSuff
ñ
Maria
Introduced because negation
fi
VSuff
la
71
Transfer to Spanish Top-Down
S
S
VP
VP
NP
NP
a
V
VSuffG
V
no
VSuffG
VSuff
N
pe
ver
VSuffG
VSuff
ñ
Maria
fi
VSuff
la
72
Transfer to Spanish Top-Down
S
S
VP
VP
NP
NP
a
V
VSuffG
V
no
VSuffG
VSuff
N
pe
ver
vi
VSuffG
VSuff
ñ
Maria
person 1 number sg mood indicative tense
past
fi
VSuff
la
73
Transfer to Spanish Top-Down
S
S
Pass features over to Spanish side
VP
VP
NP
NP
a
V
VSuffG
V
no
VSuffG
VSuff
N
pe
vi
N
VSuffG
VSuff
ñ
Maria
María
fi
VSuff
la
74
I Didnt see Maria
S
S
VP
VP
NP
NP
a
V
VSuffG
V
no
VSuffG
VSuff
N
pe
vi
N
VSuffG
VSuff
ñ
Maria
María
fi
VSuff
la
75
(No Transcript)