Underspecifying and Predicting Voice for Surface Realisation Ranking

What it is about

The authors regenerate German corpus sentences with a reversible LFG grammar and a ranker that chooses among the candidate realisations. The candidates are made to include both active and passive variants. The key step is the input: if the semantic representation keeps the original voice, the candidate set contains only one voice. The authors therefore underspecify the voice in two ways. They remove the syntactic function of the subject and object, and they add heuristics that treat a passive with an implicit agent as a one-role transitive, and a set of "trivial" agent pronouns (generic "one", universal quantifiers) and causal prepositions (by means of) as matching their active counterparts. The naive representation would make some sentences impossible to realise in the other voice, so the heuristics are needed. The authors admit that the heuristics add noise. With the heuristic representation and the ranker, the voice is predicted correctly in 91.5% of cases against a baseline of 82.5%. A pilot human evaluation on the voice data is reported alongside.

Ideas

• I1. Underspecify voice in the meta-world. The meta token should carry no active or passive; the encoder chooses by a counted table keyed by the status of the agent (present or implicit), the pronoun class and the preposition. Measure the voice agreement with the source on KJV and WEB, against the most-frequent-voice baseline the paper also reports.
• I2. Keep the trivial-agent and causal-preposition sets as BTreeSet data, not code. A passive with an implicit agent then matches its active form in the meta-world, and the gain can be counted: how many verses change their meta-world after the rule.
• I3. Score candidate voices with counts, not a learned ranker. A count model over (verb concept, agent class, voice) is deterministic and explains each choice; compare it with a single ranker score only after the baseline is reported.
• I4. Guard against the noise the authors admit. A voice delta for an author should pass Mova's existing rule (at least three more tokens fixed, holding in half of contexts). Run that rule on a voice-only delta for WEB and report whether it survives.

For Mova

The encoder in tools/translate/src/profile.rs chooses voice implicitly today. The ideas touch open problems 2 (noisy merges, here the heuristic sets) and 8 (style deltas per class). The paper's own numbers are for German and cannot be carried over; only the design of the underspecified input transfers.

Topics: Machine translation

Cite

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

@inproceedings{zarries2011underspecifying,
  title = {{Underspecifying and Predicting Voice for Surface Realisation Ranking}},
  author = {Zarrieß, Sina and Cahill, Aoife and Kuhn, Jonas},
  booktitle = {Proceedings of the 49th Annual Meeting of the Association for Computational Linguistics: Human Language Technologies},
  year = {2011},
  url = {https://aclanthology.org/P11-1101/}
}

References

33 references, 4 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Judith Aissen. 1999. Markedness and subject choice in optimality theory. Natural Language and Linguistic Theory, 17(4):673–711. find
2. Judith Aissen. 2003. Differential Object Marking: Iconicity vs. Economy. Natural Language and Linguistic Theory, 21:435–483. find
3. Anja Belz and Eric Kow. 2010. Comparing rating scales and preference judgements in language evaluation. In Proceedings of the 6th International Natural Language Generation Conference (INLG’10). find
4. Anja Belz, Mike White, Josef van Genabith, Deirdre Hogan, and Amanda Stent. 2010. Finding common ground: Towards a surface realisation shared task. In Proceedings of the 6th International Natural Language Generation Conference (INLG’10). find
5. Anja Belz. 2005. Statistical generation: Three methods compared and evaluated. In Proceedings of Tenth European Workshop on Natural Language Generation (ENLG-05), pages 15–23. find
6. Bernd Bohnet, Leo Wanner, Simon Mill, and Alicia Burga. 2010. Broad coverage multilingual deep sentence generation with a stochastic multi-level realizer. In Proceedings of the 23rd International Conference on Computational Linguistics (COLING 2010), Beijing, China. find our card
7. Joan Bresnan, Shipra Dingare, and Christopher D. Manning. 2001. Soft Constraints Mirror Hard Constraints: Voice and Person in English and Lummi. In Proceedings of the LFG ’01 Conference. find
8. Joan Bresnan, Anna Cueni, Tatiana Nikitina, and Harald Baayen. 2007. Predicting the Dative Alternation. In G. Boume, I. Kraemer, and J. Zwarts, editors, Cognitive Foundations of Interpretation. Amsterdam: Royal Netherlands Academy of Science. find
9. Aoife Cahill and Martin Forst. 2010. Human Evaluation of a German Surface Realisation Ranker. In Proceedings of the 12th Conference of the European Chapter of the ACL (EACL 2009), pages 112 – 120, Athens, Greece. Association for Computational Linguistics. find
10. Aoife Cahill and Arndt Riester. 2009. Incorporating Information Status into Generation Ranking. In Proceedings of the 47th Annual Meeting of the ACL, pages 817–825, Suntec, Singapore, August. Association for Computational Linguistics. find our card
11. Aoife Cahill, Martin Forst, and Christian Rohrer. 2007. Stochastic realisation ranking for a free word order language. In Proceedings of the Eleventh European Workshop on Natural Language Generation, pages 17–24, Saarbrücken, Germany, June. DFKI GmbH. Document D-07-01. find
12. Dick Crouch and Tracy Holloway King. 2006. Semantics via F-Structure Rewriting. In Miriam Butt and Tracy Holloway King, editors, Proceedings of the LFG06 Conference. find
13. Dick Crouch, Mary Dalrymple, Ron Kaplan, Tracy King, John Maxwell, and Paula Newman. 2006. XLE Documentation. Technical report, Palo Alto Research Center, CA. find
14. Katrin Erk and Diana McCarthy. 2009. Graded Word Sense Assignment. In Proceedings of the 2009 Conference on Empirical Methods in Natural Language Processing, pages 440 – 449, Singapore. find
15. Katja Filippova and Michael Strube. 2007. Generating constituent order in German clauses. In Proceedings of the 45th Annual Meeting of the Association for Computational Linguistics (ACL 07), Prague, Czech Republic. find
16. Katja Filippova and Michael Strube. 2009. Tree linearization in English: Improving language model based approaches. In Companion Volume to the Proceedings of Human Language Technologies Conference of the North American Chapter of the Association for Computational Linguistics (NAACL-HLT 09, short)., Boulder, Colorado. find our card
17. Martin Forst. 2007. Filling Statistics with Linguistics – Property Design for the Disambiguation of German LFG Parses. In ACL 2007 Workshop on Deep Linguistic Processing, pages 17–24, Prague, Czech Republic, June. Association for Computational Linguistics. find
18. Matthew Gerber and Joyce Chai. 2010. Beyond nombank: A study of implicit argumentation for nominal predicates. In Proceedings of the ACM Conference on Knowledge Discovery and Data Mining (KDD). find
19. Barbara J. Grosz, Aravind Joshi, and Scott Weinstein. 1995. Centering: A framework for modeling the local coherence of discourse. Computational Linguistics, 21(2):203–225. find our card
20. Thorsten Joachims. 1996. Training linear svms in linear time. In M. Butt and T. H. King, editors, Proceedings of the ACM Conference on Knowledge Discovery and Data Mining (KDD), CSLI Proceedings Online. find
21. Jonas Kuhn. 2003. Optimality-Theoretic Syntax—A Declarative Approach. CSLI Publications, Stanford, CA. find
22. Irene Langkilde and Kevin Knight. 1998. Generation that exploits corpus-based statistical knowledge. In Proceedings of the ACL/COLING-98, pages 704–710, Montreal, Quebec. find
23. Tomasz Marciniak and Michael Strube. 2005. Using an annotated corpus as a knowledge source for language generation. In Proceedings of Workshop on Using Corpora for Natural Language Generation, pages 19–24, Birmingham, UK. find
24. Hiroko Nakanishi, Yusuke Miyao, and Junichi Tsujii. 2005. Probabilistic models for disambiguation of an HPSG-based chart generator. In Proceedings of IWPT 2005. find
25. Adwait Ratnaparkhi. 2000. Trainable methods for surface natural language generation. In Proceedings of NAACL 2000, pages 194–201, Seattle, WA. find
26. Stefan Riezler, Detlef Prescher, Jonas Kuhn, and Mark Johnson. 2000. Lexicalized stochastic modeling of constraint-based grammars using log-linear measures and EM training. In Proceedings of the 38th Annual Meeting of the Association for Computational Linguistics (ACL’00), Hong Kong, pages 480–487. find
27. Stefan Riezler, Dick Crouch, Ron Kaplan, Tracy King, John Maxwell, and Mark Johnson. 2002. Parsing the Wall Street Journal using a Lexical-Functional Grammar and discriminative estimation techniques. In Proceedings of the 40th Annual Meeting of the Association for Computational Linguistics (ACL’02), Pennsylvania, Philadelphia. find
28. Eric K. Ringger, Michael Gamon, Robert C. Moore, David Rojas, Martine Smets, and Simon Corston-Oliver. 2004. Linguistically Informed Statistical Models of Constituent Structure for Ordering in Sentence Realization. In Proceedings of the 2004 International Conference on Computational Linguistics, Geneva, Switzerland. find
29. Christian Rohrer and Martin Forst. 2006. Improving coverage and parsing quality of a large-scale LFG for German. In Proceedings of LREC-2006. find
30. Erik Velldal and Stephan Oepen. 2006. Statistical ranking in tactical generation. In Proceedings of the 2006 Conference on Empirical Methods in Natural Language Processing, Sydney, Australia. find
31. Erik Velldal. 2008. Empirical Realization Ranking. Ph.D. thesis, University of Oslo, Department of Informatics. find
32. Sina Zarrieß and Jonas Kuhn. 2010. Reversing F-structure Rewriting for Generation from Meaning Representations. In Proceedings of the LFG10 Conference, Ottawa. find
33. Sina Zarrieß. 2009. Developing German Semantics on the basis of Parallel LFG Grammars. In Proceedings of the 2009 Workshop on Grammar Engineering Across Frameworks (GEAF 2009), pages 10–18, Suntec, Singapore, August. Association for Computational Linguistics. find

This page in Markdown → https://md.mova.elamur.ai/free/en/docs/papers/zarrie-2011-underspecifying-predicting-voice-surface-realisation/