Joint Hebrew Segmentation and Parsing using a PCFGLA Lattice Parser

What it is about

Hebrew function words are written attached to the next word, so one space-delimited token can have several segmentations, and a pipeline that segments first cannot use the syntax to choose. The authors put the whole segmentation into a lattice: each token becomes the set of its analyses from the MILA lexicon-based analyser, and a CKY parser reads the sentence lattice. They combine this with the Berkeley PCFG-with-latent-annotations parser, trained on about 5,240 trees (sentences 484 to 5724) and tested on 483 sentences. Parsing raw unsegmented text gives F1 79.9 (precision 82.3, recall 77.6), against 75.2 and 77.3 for a segment-then-parse pipeline and 73.7 for the best earlier result on raw text. This is an error reduction of about 20 percent. Two changes made the difference: splitting verbs into finite, non-finite and modal, and marking subject NPs. Marking object NPs did not help. Segmentation and tagging F1 is 93.8. The model still does not learn gender or number agreement.

Ideas

• I1. Joint segmentation and parsing as a Mova measure: run tools/hbo in a mode where all segmentations from its tables stay as arcs, and compare its one-best pipeline with the lattice reading on a dev split of treebanks/hbo/ptnk. Report both numbers and the share of tokens that change segmentation.
• I2. Restrict each segment to the tags its lexicon entry licenses, as the paper does. In tools/hbo this is a deterministic check on the lex and pre tables: a segment is kept only with a tag its table entry allows.
• I3. Split verb and subject labels in the treebank conversion before training, as the paper shows is the largest single gain. A Mova ablation: convert the MACULA trees with and without finite/non-finite/modal and subject marking, and compare segmentation accuracy.

For Mova

I1 and I2 belong in tools/hbo (joint mode; the segmenter tables lex and pre). I3 belongs in the conversion step from MACULA for hbo and in Syntax: the clause and the sentence as a note on verb classes. The paper is on Modern Hebrew; the lattice method transfers, the numbers do not.

Cite

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

@inproceedings{goldberg2011joint,
  title = {{Joint Hebrew Segmentation and Parsing using a PCFGLA Lattice Parser}},
  author = {Goldberg, Yoav and Elhadad, Michael},
  booktitle = {Proceedings of the 49th Annual Meeting of the Association for Computational Linguistics: Human Language Technologies},
  year = {2011},
  url = {https://aclanthology.org/P11-2124/}
}

References

20 references, 5 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Meni Adler, Yoav Goldberg, David Gabay, and Michael Elhadad. 2008. Unsupervised lexicon-based resolution of unknown words for full morphological analysis. In Proc. of ACL. find
2. Shu Cai, David Chiang, and Yoav Goldberg. 2011. Language-independent parsing with empty elements. In Proc. of ACL (short-paper). find
3. Marie Candito, Benoit Crabbé, and Djamé Seddah. 2009. On statistical parsing of French with supervised and semi-supervised strategies. In EACL 2009 Workshop Grammatical inference for Computational Linguistics, Athens, Greece. find
4. J. Chappelier, M. Rajman, R. Aragues, and A. Rozenknop. 1999. Lattice Parsing for Speech Recognition. In In Sixth Conference sur le Traitement Automatique du Langage Naturel (TANL99), pages 95–104. find
5. Yoav Goldberg and Reut Tsarfaty. 2008. A single generative model for joint morphological segmentation and syntactic parsing. In Proc. of ACL. find
6. Yoav Goldberg, Meni Adler, and Michael Elhadad. 2008. EM Can find pretty good HMM POS-Taggers (when given a good start). In Proc. of ACL. find
7. Yoav Golderg, Reut Tsarfaty, Meni Adler, and Michael Elhadad. 2009. Enhancing unlexicalized parsing performance using a wide coverage lexicon, fuzzy tag-set mapping, and em-hmm-based lexical probabilities. In Proc. of EACL. find
8. Spence Green and Christopher Manning. 2010. Better Arabic parsing: Baselines, evaluations, and analysis. In Proc. of COLING. find our card
9. Noemie Guthmann, Yuval Krymolowski, Adi Milea, and Yoad Winter. 2009. Automatic annotation of morphosyntactic dependencies in a Modern Hebrew Treebank. In Proc. of TLT. find
10. Zhongqiang Huang and Mary Harper. 2009. Self-training PCFG grammars with latent annotations across languages. In Proc. of the EMNLP, pages 832– 841. Association for Computational Linguistics. find
11. Alon Itai and Shuly Wintner. 2008. Language resources for Hebrew. Language Resources and Evaluation, 42(1):75–98, March. find
12. Dan Klein and Christopher D. Manning. 2003. Accurate unlexicalized parsing. In Proc. of ACL, Sapporo, Japan, July. Association for Computational Linguistics. find our card
13. Takuya Matsuzaki, Yusuke Miyao, and Jun’ichi Tsujii. 2005. Probabilistic CFG with latent annotations. In Proc of ACL. find
14. Slav Petrov and Dan Klein. 2008. Parsing German with latent variable grammars. In Proceedings of the ACL Workshop on Parsing German. find
15. Slav Petrov, Leon Barrett, Romain Thibaux, and Dan Klein. 2006. Learning accurate, compact, and interpretable tree annotation. In Proc. of ACL, Sydney, Australia. find our card
16. Slav Petrov. 2009. Coarse-to-Fine Natural Language Processing. Ph.D. thesis, University of California at Bekeley, Berkeley, CA, USA. find
17. Detlef Prescher. 2005. Inducing head-driven PCFGs with latent heads: Refining a tree-bank grammar for parsing. In Proc. of ECML. find
18. Khalil Sima’an, Alon Itai, Yoad Winter, Alon Altman, and Noa Nativ. 2001. Building a Tree-Bank of Modern Hebrew text. Traitement Automatique des Langues, 42(2). find
19. Reut Tsarfaty and Khalil Sima’an. 2010. Modeling morphosyntactic agreement in constituency-based parsing of Modern Hebrew. In Proceedings of the NAACL/HLT Workshop on Statistical Parsing of Morphologically Rich Languages (SPMRL 2010), Los Angeles, CA. find our card
20. Reut Tsarfaty. 2006. Integrated Morphological and Syntactic Disambiguation for Modern Hebrew. In Proc. of ACL-SRW. find our card

This page in Markdown → https://md.mova.elamur.ai/free/en/docs/papers/aclbulk-2011-goldberg-joint-hebrew-segmentation-parsing-pcfgla/