Fully Parsing the Penn Treebank

What it is about

The authors build a two-stage parser that recovers full Penn Treebank trees, including function tags and empty categories, which earlier parsers left out. The first stage is the Collins parser with a few changes (under ten lines of code) that add function tags; its Parseval accuracy matches the base parser despite the finer labels. The second stage recovers empty categories such as traces and dropped subjects with a rich feature set and a machine learning classifier, and it reduces the error on wh-traces by 41% over Levy and Manning. The paper argues that correcting parser output is much less laborious than annotating trees from scratch. Results are F1 scores on section 23 of the treebank.

Ideas

• I1. Empty-node recovery for ellipsis in prag: recover the dropped subject of a controlled or coordinated clause as an empty node, and measure the recall on the GUM enhanced empty nodes (test only, since GUM is NC).
• I2. Function-tag style second stage: after en parses a sentence, label each dependent with a grammatical group (syntactic, semantic, direction, temporal, as in the paper's Table 1) with a small classifier on the UD relation and the lexicon, and measure F1 per group on PTB section 23 (test only).
• I3. Wh-trace links for world: resolve a relative pronoun to its antecedent as a gap filler, and measure the antecedent accuracy on EWT test sentences with relative clauses (EWT is CC BY-SA, so this is a training-side measure).

For Mova

The empty-node idea touches the event layer in prag and world, and the ellipsis and word-order chapter 29. Ellipsis, word order and information structure. The relative-clause links touch 25. Relative clauses. Penn Treebank is LDC-licensed: use it for tests and measurements only, never in a shipped model. Measure first: the antecedent accuracy of relative pronouns on EWT.

Topics: Treebanks

Cite

The paper itself, not this card. BibTeX from the library's record of the work. DOI 10.3115/1220835.1220859

@inproceedings{gabbard2006fully,
  title = {{Fully Parsing the Penn Treebank}},
  author = {Gabbard, Ryan and Kulick, Seth and Marcus, Mitchell},
  booktitle = {Proceedings of the Human Language Technology Conference of the NAACL, Main Conference},
  year = {2006},
  doi = {10.3115/1220835.1220859},
  url = {https://doi.org/10.3115/1220835.1220859}
}

References

17 references, 1 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Ann Bies, Mark Ferguson, Karen Katz, and Robert Mac-Intyre. 1995. Bracketing guidelines for Treebank II style Penn Treebank project. Technical report, University of Pennsylvania. find
2. Daniel M. Bikel. 2004. On the Parameter Space of Lexicalized Statistical Parsing Models. Ph.D. thesis, Department of Computer and Information Sciences, University of Pennsylvania. find
3. E. Black, S. Abney, D. Flickinger, C. Gdaniec, R. Grishman, P. Harrison, D. Hindle, R. Ingria, F. Jelinek, J. Klavans, M. Liberman, M. Marcus, S. Roukos, B. Santorini, and T. Strzalkowski. 1991. A procedure for quantitatively comparing the syntactic coverage of English grammars. In Proceedings of the Fourth DARPA Workshop on Speech and Natural Language, pages 306–311, CA. find
4. Don Blaheta and Eugene Charniak. 2000. Assigning function tags to parsed text. In Proceedings of the 1st Annual Meeting of the North American Chapter of the Association for Computational Linguistics, pages 234–240, Seattle. find
5. Don Blaheta. 2003. Function Tagging. Ph.D. thesis, Brown University. find
6. Richard Campbell. 2004. Using linguistic principles to recover empty categories. In Proceedings of ACL. find
7. Eugene Charniak. 2000. A maximum-entropy-inspired parser. In Proceedings of the 1st Annual Meeting of the North American Chapter of the Association for Computational Linguistics. find
8. Michael Collins. 1997. Three generative, lexicalized models for statistical parsing. In Proceedings of the 35th Annual Meeting of the Association for Computational Linguistics, Madrid. find
9. Michael Collins. 2003. Head-driven statistical models for natural language parsing. Computational Linguistics, 29:589–637. find our card
10. Peter Dienes and Amit Dubey. 2003a. Antecedent recovery: Experiments with a trace tagger. In Proceedings of EMNLP. find
11. Peter Dienes and Amit Dubey. 2003b. Deep processing by combining shallow methods. In Proceedings of ACL. find
12. James Henderson. 2003. Inducing history representations for broad coverage statistical parsing. In Proceedings of NLT-NAACL 2003, Edmonton, Alberta, Canada. Association for Computational Linguistics. find
13. Valentin Jijkoun and Maarten de Rijke. 2004. Enriching the output of a parser using memory-based learning. In Proceedings of the 42nd Annual Meeting of the Association for Computational Linguistics, Barcelona, Spain. find
14. Mark Johnson. 2002. A simple pattern-matching algorithm for recovering empty nodes and their antecedents. In Proceedings of the 40st Annual Meeting of the Association for Computational Linguistics, Philadelphia, PA. find
15. Roger Levy and Christopher Manning. 2004. Deep dependencies from context–free statistical parsers: correcting the surface dependency approximation. In Proceedings of the ACL. find
16. Andrew Kachites McCallum. 2002. Mallet: A machine learning for language toolkit. http://mallet.cs.umass.edu. ↗
17. Gabriele Musillo and Paolo Merlo. 2005. Lexical and structural biases for function parsing. In Proceedings of the Ninth International Workshop on Parsing Technology, pages 83–92, Vancouver, British Columbia, October. Association for Computational Linguistics. find

This page in Markdown → https://md.mova.elamur.ai/free/en/docs/papers/bank-2006-gabbard-fully-parsing-penn-treebank/