Pith. sign in

REVIEW 3 major objections 5 minor 43 references

A Discriminative Neural Model for Cross-Lingual Word Alignment

T0 review · 3 major / 5 minor · reviewed 2026-08-14 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read A discriminative alignment module trained on a few thousand gold labels beats FastAlign and attention by 11 to 27 F1, and the better alignments lift projected Chinese NER.

desk verdict Solid, useful supervised alignment paper with a real contribution, but the headline F1 gains rest on GALE gold labels that the authors themselves admit are noisy, so the numbers should be read with caution. read the letter →

arxiv 1909.00444 v1 pith:H6KCBEES submitted 2019-09-01 cs.CL

Elias Stengel-Eskin , Tzu-Ray Su , Matt Post , Benjamin Van Durme This is my paper · ORCID
classification cs.CL
keywords wordalignmentdiscriminativeneuralmachinetranslationannotationprojectionnamedentityrecognitionlow-resourceNLPTransformer
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper argues that word alignment can be recovered as a first-class supervised task inside a neural machine translation model, rather than read off from attention. With only 1,687 Arabic and 4,871 Chinese human-annotated sentence pairs, their discriminative alignment module reaches 82.74 F1 on English-Arabic and 73.41 F1 on English-Chinese test sets, beating FastAlign by 27 and 11 points, respectively, and far exceeding averaged Transformer attention. The paper then shows these gains are not just intrinsic: Chinese NER trained on tags projected with the better alignments scores about 14 points higher than tags projected with FastAlign. This matters because alignment is the bottleneck for cross-lingual dataset projection, which lets speakers of high-resource languages build labeled data in low-resource languages.

What carries the argument

The central object is a trainable alignment matrix built from cross-lingual contextual representations: source and target states are mapped through a shared three-layer tanh network into a common space, and their dot-product similarity gives an N by M link matrix. A learned 3x3 convolution over this matrix conditions each alignment decision on its neighbors; the convolved scores pass through a sigmoid and are trained with binary cross-entropy against gold alignments. Without the convolution, F1 stays below 50, so the convolution is what turns independent pairwise similarities into coherent alignment decisions.

What would settle it

Re-annotate a random sample of the GALE Chinese and Arabic test sentences with independent, strictly protocol-checked annotators and compare DiscAlign, FastAlign, and attention on those cleaned labels; if the reported 11 to 27 point F1 advantage shrinks or disappears, the gains are partly artifacts of learning noisy gold labels rather than true alignment quality.

Watch

Extended reading notes

Core claim

The paper's model, DiscAlign, treats every source-word and target-word pair as a binary classification decision, supervised directly by gold alignments. Encoder and decoder hidden states of a pretrained Transformer are projected into a shared space with a three-layer feed-forward network; their dot products form an alignment score matrix; a 3x3 convolution makes each decision depend on neighboring alignment decisions; and a sigmoid converts the scores to probabilities. The model is pretrained on unlabeled bitext and fine-tuned on the gold alignments. On GALE test data, it scores 73.41 F1 on Chinese and 82.74 F1 on Arabic, versus 62.02 and 56.11 for FastAlign and 40.65 and 13.42 for averaged attention. Projecting English OntoNotes NER tags through the model's alignments yields a Chinese NER model at 51.76 F1 versus 37.77 with FastAlign projection.

Load-bearing premise

The load-bearing assumption is that the GALE gold alignments used for both training and evaluation are accurate enough to count as ground truth, even though the paper itself reports systematic deviations from the annotation protocol and formatting errors in that data.

Editorial extensions

If this is right

  • Even roughly 5,000 labeled alignments can beat unsupervised alignment by double digits in typologically divergent language pairs, with F1 gains of 11 for Chinese and 27 for Arabic.
  • Higher alignment quality transfers to downstream projection: projected Chinese NER improves by about 14 F1 over FastAlign projection, and projected data helps most when gold data is scarce, such as only 500 sentences.
  • Labeled alignment data matters more than unlabeled bitext: halving labeled data costs 9.44 F1, while halving pretraining bitext costs only 3.69 F1.
  • Because the alignment module is architecture-neutral and works with subword units, it can be applied to other encoder-decoder architectures, and its decision threshold can tune precision versus recall for different tasks.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Editorial inference: if the gold alignments are systematically noisy, part of the measured gain may be the model learning the annotators' protocol, including its deviations; re-annotating a test sample under a stricter protocol would isolate true alignment quality.
  • Editorial inference: the authors' focus on NER spans is a special case, and the same projection pipeline is task-agnostic, so part-of-speech tags, semantic roles, or coreference labels could be projected with comparable expected gains.
  • Editorial inference: the large impact of labeled data over bitext suggests that collecting a few thousand alignment labels for a new language pair may be more cost-effective than adding millions of parallel sentences.
  • Editorial inference: because untrained L2 speakers reached near-system F1 on the same sentences, a modest amount of annotator training on recall could plausibly make human-elicited alignments a scalable training resource.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. The paper introduces DiscAlign, a supervised discriminative alignment module integrated into a Transformer-based NMT system. The model projects contextual encoder and decoder states into a shared space via a three-layer MLP, computes pairwise dot-product similarity, applies a 3x3 convolution over the resulting alignment matrix, and trains each source-target pair with binary cross entropy against gold alignments. It is initialized from a pretrained MT model and finetuned on 1,687 Arabic and 4,871 Chinese annotated sentence pairs from GALE. The authors report intrinsic F1 gains of roughly 11-27 points over FastAlign and averaged-attention baselines on English-Chinese and English-Arabic test sets, an extrinsic English-to-Chinese NER projection experiment in which DiscAlign improves projected-data NER F1 by about 14 points over FastAlign, and a small human annotation study in which untrained L2 Chinese speakers annotate alignment at 4.4 sentences per minute with F1 comparable to DiscAlign.

Significance. If the empirical claims are reliable, this is a useful and practical contribution. The paper shows that a relatively simple supervised alignment layer, trained on only a few thousand labeled sentence pairs, can substantially outperform unsupervised aligners, and it validates the improvement on a downstream NER projection task. The two-language evaluation, the ablation showing that labeled alignment data matters more than additional unlabeled bitext, and the explicit human annotation feasibility study are notable strengths. The authors are also transparent about known quality problems in the GALE data. The main weaknesses are that the intrinsic evaluation relies on the same GALE labels used for training and that no uncertainty quantification is provided for the headline differences; these need to be addressed before the central claims can be fully accepted.

major comments (3)
  1. [Section 4.2, footnote 5; Tables 2 and 3] The GALE alignments used both to train DiscAlign and to compute every intrinsic F1 score are acknowledged to contain mismatched bitext, formatting errors, and several systematic deviations from the alignment protocol by GALE annotators. Because FastAlign and averaged attention never see these labels during training, a supervised model can inflate its reported F1 by learning annotator-specific regularities rather than true alignment quality. The human annotation experiment partially mitigates this concern, but those annotators were not trained on the GALE protocol and were evaluated against the same GALE gold labels, so it does not establish that the systematic deviations are harmless. I would like to see an evaluation on a small independently re-annotated clean test set, or an error analysis isolating the protocol-deviation cases, to demonstrate that the 11-27 point gains are not partly an artifact of fitting label noise.
  2. [Section 4.3, Tables 2 and 3] The headline F1 improvements rest on single point estimates with no variance information and no significance testing. The decision thresholds alpha are tuned on a 100-sentence validation subset, and the paper does not report the number of random restarts or seeds used. Given the small test sets (315 and 636 sentences), some of the reported gaps could be within noise. Please report confidence intervals via bootstrapping or across multiple seeds, and provide significance tests for the comparisons against FastAlign.
  3. [Section 6, Table 7] The feasibility claim for manual alignment elicitation rests on only six annotators, with average scores reported but no per-annotator breakdown, no inter-annotator agreement, and no indication of how many sentences were annotated or how sentence lengths varied across the two partitions. Since this experiment is used in the conclusion to support a call for further annotation efforts, please report annotator-level precision/recall/F1, agreement statistics, and the amount of data annotated.
minor comments (5)
  1. [Section 3, loss function] The binary cross entropy loss is written as a sum over i<N, j<M, which excludes the last source and target positions; the intended range is presumably i<=N and j<=M, and the notation should be corrected.
  2. [Figure 2(b)] The precision-recall curves are presented without threshold values on the axes or a legend, making it difficult for the reader to connect specific alpha values to the plotted operating points.
  3. [Section 4.2] The Arabic data is described only as sourced from 'local resources'; specifying the corpus or release would improve reproducibility.
  4. [Section 5] The sentence 'The data was pre-processed in following the same procedure as Section 4' is missing a word; it should read 'pre-processed following the same procedure'.
  5. [Section 6] The annotation speed of 4.4 sentences per minute is reported without a measure of sentence length or token count, which makes the speed figure difficult to interpret across datasets.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the alignment model is trained on gold-labeled GALE data and evaluated on a held-out GALE test split, with independent extrinsic NER and human annotation experiments.

full rationale

The paper's central derivation chain is not circular. DiscAlign is trained with a binary cross-entropy loss against gold alignments (Section 3) and then evaluated on held-out GALE test splits (Tables 2 and 3). This is standard supervised training/evaluation with disjoint splits, not a case of predicting a fitted quantity. The binarization threshold is tuned on a 100-sentence validation subset, but it does not determine the learned alignment function and is therefore not a fitted input masquerading as a prediction. The extrinsic NER projection experiment (Section 5) evaluates alignments through an independent downstream task with OntoNotes gold NER labels, providing evidence that is not tied to the alignment training objective. The human annotation study (Section 6) further compares DiscAlign against untrained L2 annotators on GALE development sentences, again using held-out labels. Footnote 5's observation that GALE annotators systematically deviated from the alignment protocol is a data-quality caveat that could affect validity, but it is not a circularity: both the model and the baselines are measured against the same external benchmark, and the concern that a supervised model might learn annotation bias is an empirical risk, not a definitional equivalence. No load-bearing argument reduces to a self-citation: the Sockeye citation is a toolkit implementation detail, and the Transformer backbone is standard external work. The derivation is therefore self-contained against external benchmarks and receives score 0.

Assumptions & free parameters 2 free parameters · 3 assumptions · 0 invented entities

The central empirical claim rests on the validity of GALE gold labels as ground truth, which the paper itself partially undermines; on the modeling assumption that projected contextual state similarities correspond to alignments; and on hand-tuned binarization thresholds. No new physical or mathematical entities are introduced.

free parameters (2)
  • Binarization threshold alpha = 0.15 (ZH DiscAlign), 0.94-0.99 (AR DiscAlign), 0.13-0.14 (ZH attention), 0.05-0.10 (AR attention), 0.5 (high-precision…
    Thresholds convert sigmoid outputs to binary alignments and were tuned on a 100-sentence validation subset; the wide spread across languages indicates high sensitivity to this fitted value.
  • BPE merge operations = 30K
    Hand-chosen subword vocabulary size; used for all BPE conditions and interacts with the test-time expansion heuristic that maps subword alignments to word alignments.
assumptions (3)
  • domain assumption GALE gold alignments are a valid ground truth for alignment quality
    Used as supervision and as test labels; the authors themselves note data quality problems and annotator deviations (Section 4.2, footnote 5).
  • domain assumption Dot-product similarity between projected encoder and decoder states reflects word-level semantic correspondence
    This is the central modeling postulate of the alignment module; it is not independently derived.
  • domain assumption Treating each alignment decision as a binary classification problem after a 3x3 convolution adequately models dependence between decisions
    The paper states the convolution is crucial but provides no ablation to support this claim; the independence assumption is relaxed only locally.

how reviews work

0 comments
Cite this review

Pith. "Pith review of A Discriminative Neural Model for Cross-Lingual Word Alignment." pith.science (2026). https://pith.science/paper/H6KCBEES

@misc{pith2026190900444,
  author       = {Pith},
  title        = {Pith review of: A Discriminative Neural Model for Cross-Lingual Word Alignment},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/H6KCBEES}},
  note         = {Machine review of arXiv:1909.00444}
}
read the original abstract

We introduce a novel discriminative word alignment model, which we integrate into a Transformer-based machine translation model. In experiments based on a small number of labeled examples (~1.7K-5K sentences) we evaluate its performance intrinsically on both English-Chinese and English-Arabic alignment, where we achieve major improvements over unsupervised baselines (11-27 F1). We evaluate the model extrinsically on data projection for Chinese NER, showing that our alignments lead to higher performance when used to project NER tags from English to Chinese. Finally, we perform an ablation analysis and an annotation experiment that jointly support the utility and feasibility of future manual alignment elicitation.

Figures

Figures reproduced from arXiv: 1909.00444 by the authors.

Figure 1
Figure 1. One application of alignments is to project [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. (a) Dev. F1 as function of training corpus size [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 4
Figure 4. Comparing F1 across different tag types when a model is trained on 500 gold NER sentences to when it [PITH_FULL_IMAGE:figures/full_fig_p008_4.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

43 extracted references · 28 canonical work pages

  1. [1]

    URL: " 'urlintro :=

    ENTRY address author booktitle chapter edition editor howpublished institution journal key month note number organization pages publisher school series title type volume year eprint doi pubmed url lastchecked label extra.label sort.label short.list INTEGERS output.state before.all mid.sentence after.sentence after.block STRINGS urlintro eprinturl eprintpr...

  2. [2]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in capitalize " " * FUNCT...

  3. [3]

    Z eljko Agi \'c , Anders Johannsen, Barbara Plank, H \'e ctor Alonso Mart \' nez, Natalie Schluter, and Anders S gaard. 2016. Multilingual projection for parsing truly low-resource languages. Transactions of the Association for Computational Linguistics, 4:301--312

    work page 2016

  4. [4]

    Tamer Alkhouli, Gabriel Bretschner, and Hermann Ney. 2018. On the alignment problem in multi-head attention-based neural machine translation. In Proceedings of the Third Conference on Machine Translation: Research Papers, pages 177--185

    work page 2018

  5. [5]

    Tamer Alkhouli, Gabriel Bretschner, Jan-Thorsten Peter, Mohammed Hethnawi, Andreas Guta, and Hermann Ney. 2016. https://doi.org/10.18653/v1/W16-2206 Alignment-based neural machine translation . In Proceedings of the First Conference on Machine Translation: Volume 1, Research Papers, pages 54--65. Association for Computational Linguistics

    work page doi:10.18653/v1/w16-2206 2016

  6. [6]

    Tamer Alkhouli and Hermann Ney. 2017. Biasing attention-based recurrent neural networks using external alignment information. In Proceedings of the Second Conference on Machine Translation, pages 108--117

    work page 2017

  7. [7]

    Philip Arthur, Graham Neubig, and Satoshi Nakamura. 2016. https://doi.org/10.18653/v1/D16-1162 Incorporating discrete translation lexicons into neural machine translation . In Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing, pages 1557--1567. Association for Computational Linguistics

    doi:10.18653/v1/d16-1162 2016

  8. [8]

    Dzmitry Bahdanau, Kyunghyun Cho, and Yoshua Bengio. 2014. Neural machine translation by jointly learning to align and translate. arXiv preprint arXiv:1409.0473

    arXiv 2014

Show all 43 references
  1. [9]

    Peter F Brown, Vincent J Della Pietra, Stephen A Della Pietra, and Robert L Mercer. 1993. The mathematics of statistical machine translation: Parameter estimation. Computational linguistics, 19(2):263--311

    1993

  2. [10]

    Wenhu Chen, Evgeny Matusov, Shahram Khadivi, and Jan-Thorsten Peter. 2016. Guided alignment training for topic-aware neural machine translation. arXiv preprint arXiv:1607.01628

    2016 arXiv

  3. [11]

    Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. 2018. Bert: Pre-training of deep bidirectional transformers for language understanding. arXiv preprint arXiv:1810.04805

    2018 arXiv

  4. [12]

    Chris Dyer, Victor Chahuneau, and Noah A Smith. 2013. A simple, fast, and effective reparameterization of IBM M odel 2. In Proceedings of the 2013 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pages 644--648

    2013

  5. [13]

    Steffen Eger, Johannes Daxenberger, Christian Stab, and Iryna Gurevych. 2018. Cross-lingual argumentation mining: Machine translation (and a bit of projection) is all you need! In Proceedings of the 27th International Conference on Computational Linguistics, pages 831--844

    2018

  6. [14]

    Alexander Fraser and Daniel Marcu. 2007. Measuring word alignment quality for statistical machine translation. Computational Linguistics, 33(3):293--303

    2007

  7. [15]

    Ruiji Fu, Bing Qin, and Ting Liu. 2014. Generating C hinese named entity data from parallel corpora. Frontiers of Computer Science, 8(4):629--641

    2014

  8. [16]

    Jonas Gehring, Michael Auli, David Grangier, Denis Yarats, and Yann N Dauphin. 2017. Convolutional sequence to sequence learning. In Proceedings of the 34th International Conference on Machine Learning-Volume 70, pages 1243--1252. JMLR. org

    2017

  9. [17]

    Hamidreza Ghader and Christof Monz. 2017. What does attention in neural machine translation pay attention to? In Proceedings of the Eighth International Joint Conference on Natural Language Processing (Volume 1: Long Papers), volume 1, pages 30--39

    2017

  10. [18]

    Felix Hieber, Tobias Domhan, Michael Denkowski, David Vilar, Artem Sokolov, Ann Clifton, and Matt Post. 2017. Sockeye: A toolkit for neural machine translation. arXiv preprint arXiv:1712.05690

    2017 arXiv

  11. [19]

    Rebecca Hwa, Philip Resnik, Amy Weinberg, Clara Cabezas, and Okan Kolak. 2005. Bootstrapping parsers via syntactic projection across parallel texts. Natural language engineering, 11(3):311--325

    2005

  12. [20]

    Nal Kalchbrenner and Phil Blunsom. 2013. Recurrent continuous translation models. In Proceedings of the 2013 Conference on Empirical Methods in Natural Language Processing, pages 1700--1709

    2013

  13. [21]

    Diederik P Kingma and Jimmy Ba. 2014. Adam: A method for stochastic optimization. arXiv preprint arXiv:1412.6980

    2014 arXiv

  14. [22]

    Philipp Koehn, Hieu Hoang, Alexandra Birch, Chris Callison-Burch, Marcello Federico, Nicola Bertoldi, Brooke Cowan, Wade Shen, Christine Moran, Richard Zens, et al. 2007. Moses: Open source toolkit for statistical machine translation. In Proceedings of the 45th annual meeting ...

    2007

  15. [23]

    Philipp Koehn and Rebecca Knowles. 2017. Six challenges for neural machine translation. In Proceedings of the First Workshop on Neural Machine Translation, pages 28--39

    2017

  16. [24]

    Jo \"e l Legrand, Michael Auli, and Ronan Collobert. 2016. https://doi.org/10.18653/v1/W16-2207 Neural network-based word alignment through score aggregation . In Proceedings of the First Conference on Machine Translation: Volume 1, Research Papers, pages 66--73. Association f...

    2016 doi

  17. [25]

    Xuansong Li, Stephen Grimes, Safa Ismael, Stephanie Strassel, Mohamed Maamouri, and Ann Bies. 2013. GALE A rabic- E nglish parallel aligned treebank -- broadcast news part 1; LDC 2013 T 14

    2013

  18. [26]

    Xuansong Li, Stephen Grimes, Stephanie Strassel, Xiaoyi Ma, Nianwen Xue, Mitch Marcus, and Ann Taylor. 2015. GALE C hinese- E nglish parallel aligned treebank--training; LDC 2015 T 06

    2015

  19. [27]

    Lemao Liu, Masao Utiyama, Andrew Finch, and Eiichiro Sumita. 2016. Neural machine translation with supervised attention. In Proceedings of COLING 2016, the 26th International Conference on Computational Linguistics: Technical Papers, pages 3093--3102

    2016

  20. [28]

    Jessica Ouyang and Kathleen McKeown. 2019. Neural network alignment for sentential paraphrases. In Proceedings of the 57th Conference of the Association for Computational Linguistics, pages 4724--4735

    2019

  21. [29]

    Sebastian Pad \'o and Mirella Lapata. 2009. Cross-lingual annotation projection for semantic roles. Journal of Artificial Intelligence Research, 36:307--340

    2009

  22. [30]

    Jan-Thorsten Peter, Arne Nix, and Hermann Ney. 2017. Generating alignments using target foresight in attention-based neural machine translation. The Prague Bulletin of Mathematical Linguistics, 108(1):27--36

    2017

  23. [31]

    Ofir Press and Lior Wolf. 2017. Using the output embedding to improve language models. In Proceedings of the 15th Conference of the European Chapter of the Association for Computational Linguistics: Volume 2, Short Papers, volume 2, pages 157--163

    2017

  24. [32]

    Ellen Riloff, Charles Schafer, and David Yarowsky. 2002. Inducing information extraction systems for new languages via cross-language projection. In Proceedings of the 19th international conference on Computational linguistics-Volume 1, pages 1--7. Association for Computationa...

    2002

  25. [33]

    Holger Schwenk. 2012. Continuous space translation models for phrase-based statistical machine translation. Proceedings of COLING 2012: Posters, pages 1071--1080

    2012

  26. [34]

    Rico Sennrich, Barry Haddow, and Alexandra Birch. 2016. Neural machine translation of rare words with subword units. In Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), volume 1, pages 1715--1725

    2016

  27. [35]

    Serge Sharoff. 2018. Language adaptation experiments via cross-lingual embeddings for related languages. In Proceedings of the Eleventh International Conference on Language Resources and Evaluation (LREC-2018)

    2018

  28. [36]

    Ilya Sutskever, Oriol Vinyals, and Quoc V Le. 2014. Sequence to sequence learning with neural networks. In Advances in neural information processing systems, pages 3104--3112

    2014

  29. [37]

    Akihiro Tamura, Taro Watanabe, and Eiichiro Sumita. 2014. Recurrent neural networks for word alignment model. In Proceedings of the 52nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), volume 1, pages 1470--1480

    2014

  30. [38]

    Nadi Tomeh. 2012. Discriminative Alignment Models For Statistical Machine Translation. Ph.D. thesis, Universit \'e Paris Sud-Paris XI

    2012

  31. [39]

    Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, ukasz Kaiser, and Illia Polosukhin. 2017. Attention is all you need. In Advances in Neural Information Processing Systems, pages 5998--6008

    2017

  32. [40]

    Ralph Weischedel, Martha Palmer, Mitchell Marcus, Eduard Hovy, Sameer Pradhan, Lance Ramshaw, Nianwen Xue, Ann Taylor, Jeff Kaufman, Michelle Franchini, et al. 2013. Ontonotes R elease 5.0 LDC 2013 T 19. Linguistic Data Consortium, Philadelphia, PA

    2013

  33. [41]

    Chenhai Xi and Rebecca Hwa. 2005. A backoff model for bootstrapping resources for non-english languages. In Proceedings of the conference on Human Language Technology and Empirical Methods in Natural Language Processing, pages 851--858. Association for Computational Linguistics

    2005

  34. [42]

    David Yarowsky, Grace Ngai, and Richard Wicentowski. 2001. Inducing multilingual text analysis tools via robust projection across aligned corpora. In Proceedings of the first international conference on Human language technology research, pages 1--8. Association for Computatio...

    2001

  35. [43]

    Thomas Zenkel, Joern Wuebker, and John DeNero. 2019. Adding interpretable attention to neural translation models improves word alignment. arXiv preprint arXiv:1901.11359

    2019 arXiv

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.