Elamur Research⌕
☰
home / papers / 2011 / Multilingual Annotation and Disambiguation of Discourse Connectives for Machine Translation

Multilingual Annotation and Disambiguation of Discourse Connectives for Machine Translation

Summary

The paper tackles ambiguous discourse connectives, which can signal different coherence relations depending on context. Machine translation systems often render them wrongly or drop them. The authors study three highly multi-functional connectives: French alors que and English since and while. Instead of asking annotators to apply fine-grained theoretical sense labels directly, they first have them perform translation spotting on aligned sentence pairs from the Europarl corpus, in both EN/FR and FR/EN directions. Annotators mark how each connective was rendered in the other language, and these observed translations guide a simplified, translation-relevant label set. This set roughly matches the second level of the PDTB hierarchy and makes manual annotation more reliable. Using these labels, they train classifiers in WEKA on surface, verbal, POS and dependency features. Accuracy reaches about 85% for since, which is comparable to or better than earlier work. The study matters because sense-labelled connectives should give statistical MT systems clearer signals for choosing the correct target-language connective.

Key points

  • Preliminary MT evidence: forcing correct French translations for three senses of while (Temporal, Contrast, Concession) by editing a Moses phrase table raised BLEU by 0.8 on a small 20-sentence test set.
  • since results: the SVM classifier reached 85.4% on the three-way task (Temporal, Causal, Temporal/Causal), versus 75.5% for the Maximum Entropy model of Miltsakaki et al. (2005), though on different datasets. On the two-way task it scored 85.7%, versus their 89.5%.
  • while results: cutting the label set from six to three improved accuracy by 8–10 points, to 60.9% (Miltsakaki et al. report 71.8%). Even the six-label setup beat the baseline (52.2% vs. 44.8%).
  • alors que results: a two-label scheme (Background vs. Contrast only) did not significantly beat the baseline. The authors take this as evidence against such a narrow label set.
  • Feature analysis: information-gain rankings show that contextual features (the preceding and following words) matter most across both languages. Verbal and tense features come next, and POS and dependency tags contribute least.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

Fields: Linguistics & languages

Topics: Discourse analysis · Machine translation

Cited by

18 papers of our library cite it (of the 140673 whose bibliographies we read); 6 of them have a page here, 12 not yet.

Cite

The paper itself, not this page. BibTeX from the library's record of the work.

@inproceedings{meyer2011multilingual,
  title = {{Multilingual Annotation and Disambiguation of Discourse Connectives for Machine Translation}},
  author = {Meyer, Thomas and Popescu-Belis, Andrei and Zufferey, Sandrine and Cartoni, Bruno},
  booktitle = {Proceedings of the SIGDIAL 2011 Conference},
  year = {2011},
  doi = {10.7892/boris.78682},
  url = {https://doi.org/10.7892/boris.78682}
}

References

23 references, 11 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Bruno Cartoni, Sandrine Zufferey, Thomas Meyer, and Andrei Popescu-Belis. 2011. How comparable are parallel corpora? Measuring the distribution of general vocabulary and connectives. In Proceedings of 4th Workshop on Building and Using Comparable Corpora, Portland, OR. find
  2. Eugene Charniak and Mark Johnson. 2005. Coarse-to-fine n-best parsing and maxent discriminative reranking. In Proceedings of ACL 2005 (43rd Annual Meeting of the ACL), pages 173–180, Ann Arbor, MI. find our card
  3. Pascal Denis and Benoı̂t Sagot. 2009. Coupling an annotated corpus and a morphosyntactic lexicon for state-of-the-art POS tagging with less human effort. In Proceedings of PACLIC 2009 (23rd Pacific Asia Conference on Language, Information and Computation), pages 110–119, Hong Kong, China. find
  4. David duVerle and Helmut Prendinger. 2009. A novel discourse parser based on support vector machine classification. In Proceedings of ACL-IJCNLP 2009 (47th Annual Meeting of the ACL and 4th International Joint Conference on NLP of the AFNLP), pages 665–673, Singapore. find our card
  5. Robert Elwell and Jason Baldridge. 2008. Discourse connective argument identification with connective specific rankers. In Proceedings of ICSC 2008 (2nd IEEE International Conference on Semantic Computing), pages 198–205, Santa Clara, CA. find
  6. Mark Hall, Eibe Frank, Geoffrey Holmes, Bernhard Pfahringer, Peter Reutemann, and Ian H. Witten. 2009. The WEKA data mining software: An update. ACM SIGKDD Explorations Newsletter, 11:10–18. find
  7. Thorsten Joachims. 1998. Text categorization with support vector machines: Learning with many relevant features. In Proceedings of ECML 1998 (10th European Conference on Machine Learning), pages 137– 142, Chemnitz, Germany. find our card
  8. Philipp Koehn, Hieu Hoang, Alexandra Birch, Chris Callison-Burch, Marcello Federico, Nicola Bertoldi, Brooke Cowan, Wade Shen, Christine Moran, Richard Zens, Chris Dyer, Ondrej Bojar, Alexandra Constantin, and Evan Herbs. 2007. Moses: Open source toolkit for statistical machine translation. In Proce… find
  9. Philipp Koehn. 2005. Europarl: A parallel corpus for statistical machine translation. In Proceedings of MT Summit X, pages 79–86, Phuket, Thailand. find
  10. Ziheng Lin, Hwee Tou Ng, and Min-Yen Kan. 2010. A PDTB-styled end-to-end discourse parser. Technical Report TRB8/10, School of Computing, National University of Singapore, Singapore. find
  11. William C. Mann and Sandra A. Thompson. 1988. Rhetorical Structure Theory: towards a functional theory of text organization. Text, 8(3):243–281. find
  12. Thomas Meyer. 2011. Disambiguating temporal-contrastive discourse connectives for machine translation. In Proceedings of ACL-HLT 2011 (49th Annual Meeting of the ACL: Human Language Technologies), Student Session, Portland, OR. find
  13. Eleni Miltsakaki, Nikhil Dinesh, Rashmi Prasad, Aravind Joshi, and Bonnie Webber. 2005. Experiments on sense annotations and sense disambiguation of discourse connectives. In Proceedings of the TLT 2005 (4th Workshop on Treebanks and Linguistic Theories), Barcelona, Spain. find
  14. Joakim Nivre. 2003. An efficient algorithm for projective dependency parsing. In Proceedings of IWPT 2008 (8th International Workshop on Parsing Technologies), pages 149–160, Tokyo, Japan. find our card
  15. Marie-Paule Péry-Woodley, Nicholas Asher, Patrice Enjalbert, Farah Benamara, Myriam Bras, Cécile Fabre, Stéphane Ferrari, Lydia-Mai Ho-Dac, Anne Le Draoulec, Yann Mathet, Philippe Muller, Laurent Prévot, Josette Rebeyrolle, Ludovic Tanguy, Marianne Vergez-Couret, Laure Vieu, and Antoine Widlöch… find
  16. Volha Petukhova and Harry Bunt. 2009. Towards a multidimensional semantics of discourse markers in spoken dialogue. In Proceedings of IWCS-8 (8th International Conference on Computational Semantics), pages 157–168, Tilburg, The Netherlands. find our card
  17. Emily Pitler and Ani Nenkova. 2009. Using syntax to disambiguate explicit discourse connectives in text. In Proceedings of ACL-IJCNLP 2009 (47th Annual Meeting of the ACL and 4th International Joint Conference on NLP of the AFNLP), Short Papers, pages 13–16, Singapore. find our card
  18. Emily Pitler, Mridhula Raghupathy, Hena Mehta, Ani Nenkova, Alan Lee, and Aravind Joshi. 2008. Easily identifiable discourse relations. In Proceedings of Coling 2008 (22nd International Conference on Computational Linguistics), Companion Volume: Posters, pages 87–90, Manchester, UK. find our card
  19. Rashmi Prasad, Nikhil Dinesh, Alan Lee, Eleni Miltsakaki, Livio Robaldo, Aravind Joshi, and Bonnie Webber. 2008. The Penn Discourse Treebank 2.0. In Proceedings of LREC 2008 (6th International Conference on Language Resources and Evaluation), pages 2961–2968, Marrakech, Morocco. find our card
  20. Charlotte Roze, Laurence Danlos, and Phillippe Muller. 2010. LEXCONN: a French lexicon of discourse connectives. In Proceedings of MAD 2010 (Multidisciplinary Approaches to Discourse), pages 114–125, Moissac, France. find our card
  21. Manfred Stede and Carla Umbach. 1998. DiMLex: a lexicon of discourse markers for text generation and understanding. In Proceedings of ACL 1998 (36th Annual Meeting of the ACL), pages 1238–1242, Montreal, Canada. find our card
  22. Ben Wellner, James Pustejovsky, Catherine Havasi, Roser Sauri, and Anna Rumshisky. 2006. Classification of discourse coherence relations: An exploratory study using multiple knowledge sources. In Proceedings of 7th SIGDIAL Workshop on Discourse and Dialogue, pages 117–125, Sydney, Australia. find our card
  23. Sárka Zikánová, Lucie Mladová, Jiřı́ Mı́rovský, and Pavlina Jı́nová. 2010. Typical cases of annotators’ disagreement in discourse annotations in Prague Dependency Treebank. In Proceedings of LREC 2010 (7th International Conference on Language Resources and Evaluation), pages 2002–2006, Vallet… find

↑ back to top

This page in Markdown →