Elamur Research⌕
☰
home / papers / 2016 / AMR Parsing with an Incremental Joint Model

AMR Parsing with an Incremental Joint Model

Summary

The paper addresses a weakness of pipelined Abstract Meaning Representation (AMR) parsers such as JAMR. These parsers first pick concepts and then link them with relations, so mistakes made in the first stage cannot be corrected later. The authors instead treat concept identification and relation identification as one joint, incremental problem.

They first design Component-Wise Beam Search (CWBS), a left-to-right decoder for relations. It grows a partial graph by connecting each new concept fragment to the preceding connected components through SHIFT and arc actions, and the final graph is guaranteed to be connected. CWBS is then embedded in a segment-based multiple-beam decoder, so that concept choices and relation choices can inform each other. Training uses a violation-fixing (max-violation) perceptron.

On LDC2013E117 and LDC2014T12, the joint model clearly beats pipelined baselines. It also matches or exceeds CAMR and a CCG-based parser without drawing on external semantic resources. This shows that joint incremental decoding is a practical route to better AMR parsing.

Key points

  • With gold concepts, an oracle CWBS search recovers the gold graphs with 99.6% F-score on LDC2013E117 training data and 99.7% on LDC2014T12. With learned weights, CWBS performs about as well as JAMR's non-incremental MSCG algorithm.
  • The joint model reaches F1 0.71 on both datasets, compared with 0.62 and 0.63 for the fixed version of JAMR.
  • In the Table 4 comparison, the model scores 0.71 F1. Competing systems score 0.70 (CAMR), 0.68 (CAMR* without extra features) and 0.66 (CCG-based parser on LDC2014T12).
  • On the full LDC2014T12 release, the model gets 0.66 F1. This equals CAMR, beats CAMR* (0.64) and JAMR (0.58), and falls just below the SMBT-based parser (0.67), which relies on external resources. The split is 10,312 training, 1,368 development and 1,371 test sentences.
  • Unlike the incremental parser of Damonte et al. (2016), this approach can handle non-projective graphs.

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: Machine translation

Cited by

30 papers of our library cite it (of the 140673 whose bibliographies we read); 3 of them have a page here, 27 not yet.

Cite

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/d16-1065

@inproceedings{zhou2016amr,
  title = {{AMR Parsing with an Incremental Joint Model}},
  author = {Zhou, Junsheng and Xu, Feiyu and Uszkoreit, Hans and Qu, Weiguang and Li, Ran and Gu, Yanhui},
  booktitle = {Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing},
  year = {2016},
  doi = {10.18653/v1/d16-1065},
  url = {https://doi.org/10.18653/v1/d16-1065}
}

References

24 references, 8 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Yoav Artzi, Kenton Lee, and Luke Zettlemoyer. 2015. Broad-coverage CCG Semantic Parsing with AMR. In Proc. of EMNLP, pages 1699–1710. find
  2. Laura Banarescu, Claire Bonial, Shu Cai, Madalina Georgescu, Kira Griffitt, Ulf Hermjakob, Kevin Knight, Philipp Koehn, Martha Palmer, , and Nathan Schneider. 2013. Abstract Meaning Representation for Sembanking. In Proc. of the Linguistic Annotation Workshop and Interoperability with Discourse. find our card
  3. Shu Cai and Kevin Knight. 2013. Smatch: an Evaluation Metric for Semantic Feature Structures. In Proc. of ACL, pages 748–752. find
  4. Michael Collins and Brian Roark. 2004. Incremental Parsing with the Perceptron Algorithm. In Proc. of ACL, pages 111–118. find our card
  5. Michael Collins. 2002. Discriminative Training Methods for Hidden Markov Models: Theory and Experiments with Perceptron algorithms. In Proc. of EMNLP, pages 1–8. find our card
  6. Michael A. Covington. 2001. A Fundamental Algorithm for Dependency Parsing. In Proc. of ACM Southeast Conference. find
  7. Marco Damonte, Shay B. Cohen, and Giorgio Satta. 2016. An Incremental Parser for Abstract Meaning Representation. arXiv preprint at arXiv:1608.06111. ↗ our card
  8. Jeffrey Flanigan, Sam Thomson, Jaime Carbonell, Chris Dyer, and Noah A. Smith. 2014. A Discriminative Graph-Based Parser for the Abstract Meaning Representation. In Proc. of ACL, pages 1426–1436. find
  9. Sahil Garg, Aram Galstyan, Ulf Hermjakob, and Daniel Marcu. 2016. Extracting Biomolecular Interactions Using Semantic Parsing of Biomedical Text. In Proc. of AAAI. find
  10. James Goodman, Andreas Vlachos, and Jason Naradowsky. 2016. Noise Reduction and Targeted Exploration in Imitation Learning for Abstract Meaning Representation Parsing. In Proc. of ACL, pages 1–11. find
  11. Jun Hatori, Takuya Matsuzaki, Yusuke Miyao, and Junichi Tsujii. 2012. Incremental Joint Approach to Word Segmentation, POS Tagging, and Dependency Parsing in Chinese. In Proc. of ACL, pages 1045– 1053. find
  12. Liang Huang, Suphan Fayong, and Yang Guo. 2012. Structured Perceptron with Inexact Search. In Proc. of HLT-NAACL, pages 142–151. find our card
  13. Ioannis Konstas, Frank Keller, Vera Demberg, and Mirella Lapata. 2014. Incremental Semantic Role Labeling with Tree Adjoining Grammar. In Proc. of EMNLP, pages 301–312. find
  14. Qi Li and Heng Ji. 2014. Incremental Joint Extraction of Entity Mentions and Relations. In Proc. of ACL, pages 402–412. find
  15. Fei Liu, Jeffrey Flanigan, Sam Thomson, Norman Sadeh, and Noah A. Smith. 2015. Toward Abstractive Summarization Using Semantic Representations. In Proc. of NAACL, pages 1086–1077. find
  16. Arindam Mitra and Chitta Baral. 2016. Addressing a Question Answering Challenge by Combining Statistical Methods with Inductive Rule Learning and Reasoning. In Proc. of AAAI. find
  17. Joakim Nivre. 2008. Algorithms for Deterministic Incremental Dependency Parsing. Computational Linguistics, 34(4):513–553. ↗ our card
  18. Xiaoman Pan, Taylor Cassidy, Ulf Hermjakob, Heng Ji, and Kevin Knight. 2015. Unsupervised Entity Linking with Abstract Meaning Representation. In Proc. of NAACL, pages 1130–1139. find
  19. Michael Pust, Ulf Hermjakob, Kevin Knight, Daniel Marcu, and Jonathan May. 2015. Parsing English into Abstract Meaning Representation Using Syntax-Based Machine Translation. In Proc. of EMNLP, pages 1143–1154. find our card
  20. Chuan Wang, Nianwen Xue, and Sameer Pradhan. 2015a. A Transition-based Algorithm for AMR Parsing. In Proc. of NAACL, pages 366–375. find
  21. Chuan Wang, Nianwen Xue, and Sameer Pradhan. 2015b. Boosting Transition-based AMR Parsing with Re-fined Actions and Auxiliary Analyzers. In Proc. of ACL, pages 857–862. find
  22. Keenon Werling, Gabor Angeli, and Christopher D. Manning. 2015. Robust Subgraph Generation Improves Abstract Meaning Representation Parsing. In Proc. of ACL, pages 982–991. find
  23. Yue Zhang and Stephen Clark. 2008a. A Tale of Two Parsers: Investigating and Combining Graph-Based And transition-Based Dependency Parsing Using Beam-search. In Proc. of EMNLP, pages 562–571. find our card
  24. Yue Zhang and Stephen Clark. 2008b. Joint Word Segmentation and POS Tagging Using a Single Perceptron. In Proc. of ACL, pages 888–896. find

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/zhou-2016-amr-parsing-incremental-joint-model/