Probabilistic Disambiguation Models for Wide-Coverage HPSG Parsing

What it is about

The paper builds log-linear disambiguation models for a wide-coverage HPSG grammar, trained on 20 sections of the Penn Treebank. The main obstacle is estimation cost. The authors compare four ways to reduce it: filtering only, a product model, a reference distribution and a feature function. The reference distribution method is the most accurate and cheapest of them. The model uses 598,326 features. Accuracy is clearly higher on sentences under 10 words, so short-sentence tests overestimate a parser. About one percent of sentences get no parse, which costs recall.

Ideas

• I1. en parse-eval: report LAS in three length bins (under 10 words, 10 to 19, 20 and more) on EWT test. This is a cheap check of the short-sentence effect the paper shows.
• I2. en/src/parse.rs: a cheap candidate filter before the perceptron, in the spirit of the reference distribution. It removes candidate heads that a unigram model rules out. Measure LAS and training seconds on EWT dev with en folds.
• I3. No-parse count: count EWT test sentences for which the parser returns no tree, and report recall losses on them separately.

For Mova

The code is en/src/parse.rs and the command is en parse-eval; the chapter is 20. Syntax: the Universal Dependencies frame. The paper's features are HPSG features, and its data are PTB-based and outside our licence list. Use the length check and the filter idea only, measured on EWT dev and test, with two identical runs.

Cite

The paper itself, not this card. BibTeX from the library's record of the work. DOI 10.3115/1219840.1219851

@inproceedings{miyao2005probabilistic,
  title = {{Probabilistic Disambiguation Models for Wide-Coverage HPSG Parsing}},
  author = {Miyao, Yusuke and Tsujii, Jun'ichi},
  booktitle = {Proceedings of the 43rd Annual Meeting of the Association for Computational Linguistics (ACL'05)},
  year = {2005},
  doi = {10.3115/1219840.1219851},
  url = {https://doi.org/10.3115/1219840.1219851}
}

References

27 references, 3 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. S. P. Abney. 1997. Stochastic attribute-value grammars. Computational Linguistics, 23(4). find
2. J. Baldridge and M. Osborne. 2003. Active learning for HPSG parse selection. In CoNLL-03. find
3. E. Charniak. 2000. A maximum-entropy-inspired parser. In Proc. NAACL-2000, pages 132–139. find
4. S. Chen and R. Rosenfeld. 1999. A Gaussian prior for smoothing maximum entropy models. Technical Report CMUCS-99-108, Carnegie Mellon University. find
5. S. Clark and J. R. Curran. 2004a. The importance of supertagging for wide-coverage CCG parsing. In Proc. COLING-04. find our card
6. S. Clark and J. R. Curran. 2004b. Parsing the WSJ using CCG and log-linear models. In Proc. 42th ACL. find
7. P. R. Cohen. 1995. Empirical Methods for Artificial Intelligence. MIT Press. find
8. M. Collins. 1999. Head-Driven Statistical Models for Natural Language Parsing. Ph.D. thesis, Univ. of Pennsylvania. find
9. S. Geman and M. Johnson. 2002. Dynamic programming for parsing and estimation of stochastic unification-based grammars. In Proc. 40th ACL. find
10. M. Johnson and S. Riezler. 2000. Exploiting auxiliary distributions in stochastic unification-based grammars. In Proc. 1st NAACL. find
11. M. Johnson, S. Geman, S. Canon, Z. Chi, and S. Riezler. 1999. Estimators for stochastic “unification-based” grammars. In Proc. ACL’99, pages 535–541. find
12. R. M. Kaplan, S. Riezler, T. H. King, J. T. Maxwell III, and A. Vasserman. 2004. Speed and accuracy in shallow and deep stochastic parsing. In Proc. HLT/NAACL’04. find
13. R. Malouf and G. van Noord. 2004. Wide coverage parsing with stochastic attribute value grammars. In Proc. IJCNLP-04 Workshop “Beyond Shallow Analyses”. find
14. R. Malouf. 2002. A comparison of algorithms for maximum entropy parameter estimation. In Proc. CoNLL-2002. find
15. M. Marcus, G. Kim, M. A. Marcinkiewicz, R. MacIntyre, A. Bies, M. Ferguson, K. Katz, and B. Schasberger. 1994. The Penn Treebank: Annotating predicate argument structure. In ARPA Human Language Technology Workshop. find our card
16. Y. Miyao and J. Tsujii. 2002. Maximum entropy estimation for feature forests. In Proc. find
17. Y. Miyao, T. Ninomiya, and J. Tsujii. 2003. Probabilistic modeling of argument structures including non-local dependencies. In Proc. find
18. Y. Miyao, T. Ninomiya, and J. Tsujii. 2004. Corpus-oriented grammar development for acquiring a Head-driven Phrase Structure Grammar from the Penn Treebank. In Proc. IJCNLP-04. find
19. J. Nocedal and S. J. Wright. 1999. Numerical Optimization. Springer. find
20. S. Oepen, D. Flickinger, J. Tsujii, and H. Uszkoreit, editors. 2002a. Collaborative Language Engineering: A Case Study in Efficient Grammar-Based Processing. CSLI Publications. find
21. S. Oepen, K. Toutanova, S. Shieber, C. Manning, D. Flickinger, and T. Brants. 2002b. The LinGO, Redwoods treebank. motivation and preliminary applications. In Proc. find our card
22. M. Osborne. 2000. Estimation of stochastic attribute-value grammar using an informative sample. In Proc. find
23. C. Pollard and I. A. Sag. 1994. Head-Driven Phrase Structure Grammar. University of Chicago Press. find
24. S. Riezler, T. H. King, R. M. Kaplan, R. Crouch, J. T. Maxwell III, and M. Johnson. 2002. Parsing the Wall Street Journal using a Lexical-Functional Grammar and discriminative estimation techniques. In Proc. 40th ACL. find
25. B. Taskar, D. Klein, M. Collins, D. Koller, and C. Manning. 2004. Max-margin parsing. In EMNLP 2004. find
26. K. Toutanova and C. D. Manning. 2002. Feature selection for a rich HPSG grammar using decision trees. In Proc. CoNLL-2002. find
27. K. Toutanova, P. Markova, and C. Manning. 2004. The leaf projection path view of parse trees: Exploring string kernels for HPSG parse selection. In EMNLP 2004. find

This page in Markdown → https://md.mova.elamur.ai/free/en/docs/papers/aclbulk-2005-miyao-probabilistic-disambiguation-models-wide-coverage/