1. Universität des Saarlandes, Saarbrücken
Lernverfahren auf Basis von Parallelkorpora Learning Techniques based on Parallel Corpora
Jonas Kuhn
Universität des Saarlandes, Saarbrücken
Heidelberg, 16. Juni 2005

2. Talk Outline Parallel Corpora
The PTOLEMAIOS Project: Goals and Motivation
Specification of the intended prototype system
Study 1
Unsupervised learning of a probabilistic context-free grammar (PCFG) using partial information from a parallel corpus
Study 2
A chart-based parsing algorithm for bilingual parallel grammars based on a word alignment
Conclusion
The PTOLEMAIOS research agenda

3. Parallel Corpora
Collections of texts and their translation into different languages
Alignment across languages at various levels
Document
Section
Paragraph
Sentence (not necessarily one-to-one)
Phrase
Word
Varying quality depending on origin of translation
Translation is often not literal
Parts of a document may occur only in one version

4. Examples of Parallel Corpora
Hansards of the 36th Parliament of Canada (
1.3 million sentence pairs (19.8 million word forms English/21.2 million word forms French)
The Bible
Europarl corpus (European Parliament Proceedings Parallel Corpus ) (
11 languages: Danish, Dutch, English, Finnish, French, German, Greek, Italian, Portuguese, Spanish, Swedish
Up to 28 million word forms per language
OPUS – an open source parallel corpus (
Includes Europarl corpus and various manuals for open source software (parts of which have been translated into more than 20 languages, incl. e.g., Chinese, Hebrew, Japanese, Korean, Russian, and Turkish)

5. Uses for Parallel Corpora
Building bilingual dictionaries
Resource for Machine Translation research
“Classical” MT: data basis for transfer rules/dictionary entries
Statistical MT: training data for statistical alignment models
Paragraph level
Word level
Resource for (multilingual) NLP applications
“Annotation projection” for training monolingual tools

6. Query interface From Opus corpus website
(based on the corpus work bench/CQP from IMS Stuttgart)

7. Query interface

8. Training Statistical Alignments
GIZA++ Tool
Implementation of “IBM models” for Statistical MT
Extension of the program GIZA (part of the SMT toolkit EGYPT) developed at a summer workshop in 1999 at Johns-Hopkins University
Sample word alignment (from Europarl corpus):
This
was
actually
referred to in
the
Minutes
Dies
war im
Protokoll
nachzulesen

9. “Annotation Projection”
Yarowsky/Ngai/Wicentowski 2001
Parallel corpus and word alignment given
Tagger/chunker for English exists
Projected annotation is used as training data for a tagger/chunker in the target language
Robust learning techniques based on confidence in training data DT JJ NN IN JJ NN [ a
significant
producer ]
for [
crude
oil ] [ un
producteur
important ] de [
petrole
brut ] DT NN JJ IN NN JJ

10. Quality of projected information
Evaluation results for part-of-speech tagging (from Yarowsky et al. 2001)

11. Train grammars on parallel corpora?
New weakly supervised learning approach for (probabilistic) syntactic grammars:
Training data: Parallel corpora – collections of original texts and their translations into one or more languages
Preparatory step: Identification of word correspondences with known statistical techniques (word alignment from statistical machine translation) en de fr
Heute
stellt
sich
die
Lage
jedoch
völlig
anders
dar
The
situation
now
however is
radically
different

12. Train grammars on parallel corpora?
Beyond lexical information, patterns in the word correspondence relation contain rich implicit information about the grammars of the languages
One should be able to exploit this implicit information about structure and meaning for grammar learning
Little manual annotation effort should be required
Combination of insights from linguistics and machine learning
Heute
stellt
sich
die
Lage
jedoch
völlig
anders
dar
The
situation
now
however is
radically
different

13. The PTOLEMAIOS Project
Parallel Corpus-Based Grammar Induction: PTOLEMAIOS
Parallel-Text-based Optimization for Language Learning – Exploiting Multilingual Alignment for the Induction Of Syntactic Grammars
Funded by DFG (German Research Foundation) as an Emmy Noether research group
Universität des Saarlandes (Saarbrücken), Department of Computational Linguistics
Starting date: 1 April 2005
Expected duration: 4 years (1-year extension possible)
Rosetta Stone

14. Project Goals
Development of formalisms and algorithms to support grammar induction for arbitrary languages from parallel corpora
To make goals tangible…
Intended prototype:
The PTOLEMAIOS I system for building grammars for new (sub-)languages

15. The PTOLEMAIOS I system
Resources required:
Parallel corpus of language L and one or (ideally) more other languages
No NLP tools for language L required
Preparatory work required:
Manual annotation of a set of seed sentence pairs (e.g., pairs)
Phrasal correspondence across languages
“Lean” bracketing: mark only
full argument/modifier phrases (PPs, NPs)
and full clauses

16. The PTOLEMAIOS I system
Training steps:
(Sentence alignment on parallel corpus)
Word alignment on parallel corpus
Using standard techniques from Statistical Machine Translation (GIZA++ tool)
Part-of-speech clustering for L
Bootstrapping learning of syntactic grammars for L and the other language(s)
Starting from annotated seed data
Exploit large amounts of unannotated data, finding systematic patterns in phrasal correspondences
Assuming implicit underlying representation (“pseudo meaning representation”)
Relying on consensus across the grammars

17. The PTOLEMAIOS I system
Result:
Robust probabilistic grammar for L
Representation of predicate-argument and modifier relations
Models predict probabilities for cross-linguistic argument/modifier links
(These will be particularly useful in lexicalized models)
Application:
Multilingual Information Extraction, Question Answering
Intermediate step for syntax-based MT

18. Motivation Practical Theoretical
Explore alternative to standard treebank training of grammars
For “smaller” languages, it is unrealistic to do the necessary manual resource annotation
Theoretical
Establish parallel corpora as an empirical basis for (crosslinguistic or monolingual) linguistic studies
Frequency-related phenomena (like multi-word expressions/collocations) are otherwise hard to assess empirically at the level of syntax
Learnability properties as a criterion for assessing formal models for natural language

19. Study 1
Unsupervised learning of a probabilistic context-free grammar (PCFG) exploiting partial information from a parallel corpus [Kuhn 2004 (ACL)]
Underlying consideration:
The distribution of word correspondences in the translation of a string contains partial information about possible phrase boundaries
Heute
stellt
sich
die
Lage
jedoch
völlig
anders
dar
The
situation
now
however is
radically
different

20. Alignment-induced word blocks
Word alignments obtained with the standard model are asymmetrical [Brown et al. 1993]
Each word from language L1 is mapped to a set of words from L2
(For L2-words without an overt correspondent in L1, an artificial word NULL is assumed in each sentence in L1)
Definition
A word alignment mapping  induces a Word block w1...wn in language L1, iff the union over (w1)...(wn) forms a continuous string in L2 [separated only by words from (NULL)]
(The)={die}
(situation)={Lage} 
(NULL)={der} L2
Heute
ist
die
Lage in
der
Türkei
anders L1
NULL
The
situation in
Turkey
now is
different

21. Word blocks and constituents
Let’s call a word block maximal if adding a word to the left or right leads to a non-word block
Maximal word blocks are possible constituents – but not necessary ones
Conservative formulation:
Exclusion of impossible constituents (“distituents”)
… whenever a word sequence crosses block boundaries …
… without fully covering one of the adjacent blocks
Wir
brauchen
bald
eine
Entscheidung We
need a
decision
soon

22. Unsupervised Learning
Grammar induction of an X-bar grammar
Using a variant of standard PCFG induction with the Inside-Outside algorithm (an Expectation Maximization algorithm)
All word spans are considered as phrase candidates – except the excluded distituents
Automatic generalization based on patterns in the learning data (after part-of-speech tagging)
Exclusion of distituents can reduce the effect of frequent non-phrasal word sequences

23. Empirical Results Comparative experiment [Kuhn 2004 (ACL)]
A: Grammar induction based on English corpus data only
B: Induction including partial information from parallel corpus (Europarl corpus) [Koehn 2002]
Statistical word alignment, trained with GIZA++
[Al-Onaizan et al., 1999; Och/Ney, 2003]
Exclusion of “distituents” based on alignment-induced word blocks
Evaluation:
Parsing sentences from the Penn Treebank (Wall Street Journal) with the induced grammars
Comparison of the “automatical” analyses with the gold standard treebank analyses created by linguists

24. Empirical Results Mean of Precision und Recall
# correctly identified phrases
# proposed phrases
# correctly identified phrases
# gold standard phrases

25. Study 2 Underlying consideration:
It should be possible to learn more complex syntactic generalization from parallel corpora if phrase correspondences can be observed systematically
Cross-linguistic “consensus structure” as a poor man’s meaning representation (pseudo-meaning representation)
Form-meaning relation is important for Optimality Theory-style discriminative learning S XP S XP XP XP XP
look
must we So
agricultural
the at
policy XP XP XP
Wir
müssen
deshalb
die
Agrarpolitik
prüfen

26. Parallel Parsing
Prerequisites for structural learning from parallel corpora:
Grammar formalism generating sentence pairs or tuples (bitexts or multitexts)
Generalization of monolingual context-free grammars, generating two or more trees simultaneously (keeping note of phrase correspondences)
Algorithms for parsing and learning (Parallel Parsing/Synchronous Parsing)
Problem: higher complexity than monolingual parsing
Compare theoretical work on machine translation [Wu 1997, Melamed 2004]
Focus for this study:
Efficient parallel parsing based on a word-alignment
[Kuhn 2005 (IJCAI)]

27. Chart Parsing (review)
In chart parsing for context-free grammars, partial analyses covering the same string under the same category are merged
WP covering string position j through m
Two possible internal analyses – a single chart entry
Internal distinction irrelevant from external point of view
WP: j-m
Reading 1
XP: j-k YP: k-m
WP  XP YP
 WP: j-m
Reading 2
ZP: j-n XP: n-m
WP  ZP XP WP YP XP XP ZP
i k n m

28. Data Structure for Parallel Parsing
The input is not a one-dimensional word string like in classical parsing, but a two-dimensional word array
Representation of the
word alignment
policy
agricultural
the at
look
must we So
agricultural
the at
policy
look
must we So
Wir
Wir
prüfen
Agrarpolitik
die
deshalb
müssen
Wir
müssen
müssen
deshalb
deshalb
die
die
Agrarpolitik
Agrarpolitik
prüfen
prüfen

29. “3-D” Parsing policy agricultural the at prüfen look Agrarpolitik must
die
deshalb
müssen
Wir XP S
prüfen
“3-D” Parsing XP
look
must we So
agricultural
the at
policy S
policy
agricultural
the at
prüfen
look
Agrarpolitik
must
die we
deshalb
müssen So
Wir

30. Assumptions for Parallel Parsing
A particular word alignment is given (or the n best alignments according to a statistical model)
Note: Both languages may contain words without a correspondent (“null words”)
Complete constituents must be continuous in both languages
There may however be phrase-internal “discontinuities”
soon
decision a
need We
Wir
brauchen
bald
eine
Entscheidung

31. Generalized Earley Parsing
Main idea:
One of the languages is the “master language”
Choice of master language is arbitrary in principle
For efficiency: pick language with more null words
Primary index for chart entries according to master language: string span (from-to)
Secondary index: Bit vector for word coverage in the secondary language (cp. chart-based generation)
Example:
Active chart entry for partial constituent
5: soon
4: decision
3: a
2: need
PI: 1-3, SI: [01001]
1: We
1:Wir
2:brauchen
3:bald
4:eine
5:Entscheidung

32. Primary/Secondary Index in Parsing
Combination of chart entries
5: soon
4: decision
3: a
2: need
1: We
1:Wir
2:brauchen
3:bald
4:eine
5:Entscheidung
PI: 1-3, SI: [01001]
PI: 3-5, SI: [00110] 
PI: 1-5, SI: [01111]

33. Illegal as a passive chart entry
Complexity
With a given word alignment, the secondary index is fully determined by the primary index unless there are any null words in the secondary language
In the absence of null words the worst case parsing complexity is essentially the same as in monolingual parsing
The secondary index does not add any free variables to the inference rules for parsing
(Bitvector operations are very cheap – ignored here)
In the average case the search space is even reduced over monolingual parsing
Illegal as a passive chart entry
Example:
(Incorrect) combination of a chart entry with an incomplete constituent is excluded early
PI: 0-1, SI: [10000]
5: soon
PI: 1-3, SI: [01001]
4: decision
3: a
2: need
PI: 0-3, SI: [11001]
1: We
1:Wir
2:brauchen
3:bald
4:eine
5:Entscheidung

34. Complexity and Null Words
Null words in the secondary language L2 create possible variations for the secondary index (with a single primary index)
However the effect is fairly local thanks to the continuity assumption
policy
agricultural
the at
look
must we So
Wir
müssen
deshalb
die
Agrarpolitik
prüfen
look
agricultural
the at
policy XP
1. PI: 3-3, SI: [0,0,0,0,1,0,0,0]
PI: 3-4, SI: [0,0,0,0,0,1,0,0]
PI: 3-4, SI: [0,0,0,0,1,1,0,0]
2. PI: 3-3, SI: [0,0,0,0,1,0,0,0]
PI: 3-5, SI: [0,0,0,0,0,1,1,1]
PI: 3-5, SI: [0,0,0,0,1,1,1,1]
3. PI: 3-3, SI: [0,0,0,0,1,0,0,0]
PI: 3-7, SI: [0,0,0,1,0,1,1,1]
 PI: 3-7, SI: [0,0,0,1,1,1,1,1]
look
agricultural
the at
policy XP
look
agricultural
the at
policy XP
8: policy
7: agricultural
6: the
5: at
4: look
4: die
5: Agrarpolitik
6: prüfen
No passive entry, i.e., can be used only locally

35. Complexity and Null Words
Variability due to null words increases the worst case complexity depending on the number of null words in L2
n: total number of words in L1
m: number of null words in L2
(note: typically clearly smaller than n)
Complexity class for alignment-based parallel parsing (time complexity for non-lexicalized grammars): O(n3m3)
For comparison – complexity for general parallel parsing without a fixed alignment: O(n6)
[vgl. Wu 1997, Melamed 2004]
O(n6)
O(n3m3)
O(n3)

36. Experimental Results Prototype implementation of parser Comparison of
SWI Prolog (chart implemented as a hash function)
Probabilistic variant: Viterbi parser (determining the most probable reading)
Scripts for generation of training data (currently based on hand-labeled data using MMAX2) [
Comparison of
“Correspondence-guided synchronous parsing” (CGSP)
(Simulated) monolingual parsing

37. Experimental Results

38. Experimental Results

39. Alignment-based Parallel Parsing
Conclusion: Alignment-based parsing of parallel corpora on a larger scale should be realistic
Use in (weakly supervised) grammar learning should be possible
Sentences with a large proportion of null words on both sides could be filtered out
Open questions:
Effective heuristic for determining a single word alignment from a statistical alignment
Tractable relaxation of continuity assumption

40. Talk Summary Parallel Corpora
The PTOLEMAIOS Project: Goals and Motivation
Specification of the intended prototype system
Study 1
Unsupervised learning of a probabilistic context-free grammar (PCFG) using partial information from a parallel corpus
Study 2
A chart-based parsing algorithm for bilingual parallel grammars based on a word alignment
Conclusion
The PTOLEMAIOS research agenda

41. The PTOLEMAIOS research agenda
Formal grammar model for parallel linguistic analysis
Specific choice of linguistic representations/constraints
Efficient parallel parsing algorithms
Probability models for parallel structural representations
Weakly supervised learning techniques for bootstrapping the grammars
Grammar
formalism
Algorithmic
realization
Linguistic
specification
Probabilistic
modeling
Bootstrapping
learning

42. Conclusion
The planned PTOLEMAIOS architecture

43. Selected References
Al-Onaizan, Yaser, Jan Curin, Michael Jahr, Kevin Knight, John Lafferty, Dan Melamed, Franz-Josef Och, David Purdy, Noah A. Smith, and David Yarowsky Statistical machine translation. Final report, JHU Workshop.
Brown, P.F., S. A. Della Pietra, V. J. Della Pietra, and R. L. Mercer The mathematics of statistical machine translation: Parameter estimation. Computational Linguistics 19:263–311.
Dubey, Amit, and Frank Keller Probabilistic parsing for German using sister-head dependencies. In Proceedings of the 41st Annual Meeting of the Association for Computational Linguistics, pp. 96–103, Sapporo.
Koehn, Philipp Europarl: A multilingual corpus for evaluation of machine translation. Ms., University of Southern California.
Kuhn, Jonas Experiments in parallel-text based grammar induction. In Proceedings of the 42nd Annual Meeting of the Association for Computational Linguistics (ACL 2004). Barcelona.
Melamed, I. Dan Multitext grammars and synchronous parsers. In Proceedings of the 42nd Annual Meeting of the Association for Computational Linguistics (ACL 2004). Barcelona.
Och, Franz Josef and Hermann Ney A systematic comparison of various statistical alignment models. Computational Linguistics, 29(1):19–51.
Wu, Dekai Stochastic inversion transduction grammars and bilingual parsing of parallel corpora. Computational Linguistics, 23(3):377–403.
Yarowsky, D., G. Ngai and R. Wicentowski Inducing Multilingual Text Analysis Tools via Robust Projection across Aligned Corpora. In Proceedings of HLT. 161—168.