Elamur Research⌕
☰
home / papers / 2018 / Zero-Shot Transfer Learning for Event Extraction

Zero-Shot Transfer Learning for Event Extraction

Summary

Supervised event extractors depend on hand-labeled examples for each event type, so covering a new type normally requires fresh annotation. Huang et al. reframe event extraction as grounding: each event mention is linked to the closest type in a target ontology, and this is the first application of zero-shot learning to the task. Candidate triggers and arguments come from AMR parses, using WSD linked to OntoNotes and FrameNet lexical units. Each mention becomes a set of AMR tuples, with one matrix per AMR relation. Each event type is represented by its name plus its role names, composed through a shared tensor. A single CNN encodes both kinds of structure into a common vector space. Training on a few seen types pulls mentions toward their gold types, and added negative instances reduce overfitting. At test time, unseen-type mentions receive the nearest type. On 23 unseen ACE types, the system matched a supervised model trained on roughly 3,000 annotated sentences, without any labels for those types. This makes event extraction scalable to new real-world scenarios.

Key points

  • Adding an unseen type needs only its ontology definition (type and role names) plus annotations for a handful of seen types. ACE's 33 types are few compared with vision ZSL label sets, which is why the authors designed a negative-sampling loss.
  • Main result: on 23 unseen types, the system performs on par with a supervised baseline trained on 3,000 sentences containing 500 event mentions.
  • In the ERE experiment, 6 seen types (Arrest-Jail, Execute, Die, Meet, Sentence, Charge-Indict; 548 mentions) plus 500 negatives were used for training. Evaluation used 200 sentences containing Attack and Convict mentions.
  • With gold AMR, trigger F1 was 59.1 and argument F1 was 23.2. Keeping only core roles (an SRL proxy) gave 58.4 and 18.2, and system AMR gave 46.7 and 18.6. This suggests that fine-grained AMR relations help argument labeling.
  • Code is released at github.com/wilburOne/ZeroShotEvent.

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: Natural language processing

Cited by

87 papers of our library cite it (of the 140884 whose bibliographies we read); 6 of them have a page here, 81 not yet.

Cite

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/p18-1201

@inproceedings{huang2018zero,
  title = {{Zero-Shot Transfer Learning for Event Extraction}},
  author = {Huang, Lifu and Ji, Heng and Cho, Kyunghyun and Dagan, Ido and Riedel, Sebastian and Voss, Clare},
  booktitle = {Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)},
  year = {2018},
  doi = {10.18653/v1/p18-1201},
  url = {https://doi.org/10.18653/v1/p18-1201}
}

References

53 references, 14 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Collin F Baker, Charles J Fillmore, and John B Lowe. 1998. The berkeley framenet project. In Proc. COL-ING1998. find our card
  2. L. Banarescu, C. Bonial, S. Cai, M. Georgescu, K. Griffitt, U. Hermjakob, K. Knight, P. Koehn, M. Palmer, and N. Schneider. 2013. Abstract meaning representation for sembanking. In Proc. ACL2013 Workshop on Linguistic Annotation and Interoperability with Discourse. find our card
  3. M. Banko, M. Cafarella, S. Soderland, M. Broadhead, and O. Etzioni. 2007. Open information extraction for the web. In Proc. IJCAI2007. find
  4. M. Banko, O. Etzioni, and T. Center. 2008. The trade-offs between open and traditional relation extraction. In Proc. ACL-HLT2008. find
  5. Ankur Bapna, Gokhan Tur, Dilek Hakkani-Tur, and Larry Heck. 2017. Towards zero-shot frame semantic parsing for domain scaling. arXiv preprint arXiv:1707.02363 . — [arXiv:1707.02363](https://arxiv.org/abs/1707.02363) find
  6. Ofer Bronstein, Ido Dagan, Qi Li, Heng Ji, and Anette Frank. 2015. Seed-based event trigger labeling: How far can event descriptions get us? In Proc. ACL2015. find
  7. Soravit Changpinyo, Wei-Lun Chao, and Fei Sha. 2017. Predicting visual exemplars of unseen classes for zero-shot learning. In Proc. ICCV2017. find
  8. Long Chen, Hanwang Zhang, Jun Xiao, Wei Liu, and Shih-Fu Chang. 2017. Zero-shot visual recognition using semantics-preserving adversarial embedding network. arXiv preprint arXiv:1712.01928 . — [arXiv:1712.01928](https://arxiv.org/abs/1712.01928) find
  9. Y. Chen, L. Xu, K. Liu, D. Zeng, and J. Zhao. 2015. Event extraction via dynamic multi-pooling convolutional neural networks. In Proc. ACL2015. find our card
  10. O. Etzioni, M. Cafarella, D. Downey, A. Popescu, T. Shaked, S. Soderland, D. Weld, and A. Yates. 2005. Unsupervised named-entity extraction from the web: An experimental study. Artificial Intelligence . find
  11. O. Etzioni, A. Fader, J. Christensen, S. Soderland, and M. Mausam. 2011. Open information extraction: The second generation. In Proc. IJCAI2011. find
  12. X. Feng, L. Huang, D. Tang, B. Qin, H. Ji, and T. Liu. 2016. A language-independent neural network for event detection. In Proc. ACL2016. find our card
  13. A. Frome, G. Corrado, J. Shlens, S. Bengio, J. Dean, and T. Mikolov. 2013. Devise: A deep visual-semantic embedding model. In Proc. NIPS2013. find
  14. Y. Ma, E. Cambria, and S. Gao. 2016. Label embedding for zero-shot fine-grained named entity typing. In Proc. COLING2016. find
  15. S. Hochreiter and J. Schmidhuber. 1997. Long short-term memory. Neural computation . find
  16. D. McClosky, M. Surdeanu, and C. D. Manning. 2011. Event extraction as dependency parsing. In Proc. ACL2011. find
  17. Y. Hong, J. Zhang, B. Ma, J. Yao, G. Zhou, and Q. Zhu. 2011. Using cross-entity inference to improve event extraction. In Proc. ACL2011. find our card
  18. L. Huang, T. Cassidy, X. Feng, H. Ji, C. Voss, J. Han, and A. Sil. 2016. Liberal event extraction and event schema induction. In Proc. ACL2016. find
  19. T. Mikolov, K. Chen, G. Corrado, and J. Dean. 2013. Efficient estimation of word representations in vector space. CoRR abs/1301.3781. find
  20. M. Miwa, R. Stre, Y. Miyao, and J. Tsujii. 2009. A rich feature vector for protein-protein interaction extraction from multiple corpora. In Proc. EMNLP2009. find
  21. L. Huang, J. May, X. Pan, H. Ji, X. Ren, J. Han, L. Zhao, and J. Hendler. 2017. Liberal entity extraction: Rapid construction of fine-grained entity typing systems. Big Data . find
  22. T. Nguyen, K. Cho, and R. Grishman. 2016. Joint event extraction via recurrent neural networks. In Proc. NAACL-HLT2016. find our card
  23. H. Ji and R. Grishman. 2008. Refining event extraction through cross-document inference. In Proc. ACL2008. find our card
  24. T. Nguyen and R. Grishman. 2015. Event detection and domain adaptation with convolutional neural networks. In Proc. ACL2015. find our card
  25. Heng Ji and Ralph Grishman. 2011. Knowledge base population: Successful approaches and challenges. In Proc. ACL-HLT2011. find
  26. M. Norouzi, T. Mikolov, S. Bengio, Y. Singer, J. Shlens, A. Frome, G. Corrado, and J. Dean. 2013. Zero-shot learning by convex combination of semantic embeddings. arXiv preprint arXiv:1312.5650 . — [arXiv:1312.5650](https://arxiv.org/abs/1312.5650) find
  27. H. Kim, X. Ren, Y. Sun, C. Wang, and J. Han. 2013. Semantic frame-based document representation for comparable corpora. In Proc. ICDM2013. find
  28. Elyor Kodirov, Tao Xiang, Zhenyong Fu, and Shaogang Gong. 2015. Unsupervised domain adaptation for zero-shot learning. In Proc. ICCV2015. find
  29. Omer Levy, Minjoon Seo, Eunsol Choi, and Luke Zettlemoyer. 2017. Zero-shot relation extraction via reading comprehension. arXiv preprint arXiv:1706.04115 . find
  30. Q. Li, H. Ji, and L. Huang. 2013. Joint event extraction via structured prediction with global features. In Proc. ACL2013. find our card
  31. Yanan Li, Donghui Wang, Huanhang Hu, Yuetan Lin, and Yueting Zhuang. 2017. Zero-shot recognition using dual visual-semantic mapping paths. arXiv preprint arXiv:1703.05002 . — [arXiv:1703.05002](https://arxiv.org/abs/1703.05002) find
  32. S. Liao and R. Grishman. 2010. Using document level cross-event inference to improve event extraction. In Proc. ACL2010. find our card
  33. B. Liu, L. Qian, H. Wang, and G. Zhou. 2010. Dependency-driven feature-based learning for extracting protein-protein interactions from biomedical text. In Proc. COLING2010. find
  34. M. Palmer, D. Gildea, and N. Xue. 2010. Semantic role labeling. Synthesis Lectures on Human Language Technologies . find
  35. Haoruo Peng, Yangqiu Song, and Dan Roth. 2016. Event detection and co-reference with minimal supervision. In Proc. EMNLP2016. find our card
  36. Kuan-Chuan Peng, Ziyan Wu, and Jan Ernst. 2017. Zero-shot deep domain adaptation. arXiv preprint arXiv:1707.01922 . — [arXiv:1707.01922](https://arxiv.org/abs/1707.01922) find
  37. J. Pustejovsky. 1991. The syntax of event structure. Cognition . find
  38. L. Qu, G. Ferraro, L. Zhou, W. Hou, and T. Baldwin. 2016. Named entity recognition for novel types by transfer learning. In Proc. ACL2016. find
  39. S. Riedel and A. McCallum. 2011. Fast and robust joint models for biomedical event extraction. In Proc. EMNLP2011. find
  40. A. Ritter, O. Etzioni, and S. Clark. 2012. Open domain event extraction from twitter. In Proc. SIGKDD2012. find
  41. Bernardino Romera-Paredes and Philip Torr. 2015. An embarrassingly simple approach to zero-shot learning. In Proc. ICML2015. find
  42. S. Liu, Y. Chen, S. He, K. Liu, and J. Zhao. 2016. Leveraging framenet to improve automatic event detection. In Proc. ACL2016. find our card
  43. S. Sekine. 2006. On-demand information extraction. In Proc. COLING-ACL2006. find
  44. Wei Lu and Dan Roth. 2012. Automatic event extraction with structured preference modeling. In Proc. ACL2012. find
  45. Y. Shinyama and S. Sekine. 2006. Preemptive information extraction using unrestricted relation discovery. In Proc. HLT-NAACL2006. tion using compositional universal schema. In Proc. NAACL2016. find
  46. R. Socher, M. Ganjoo, C. Manning, and A. Ng. 2013a. Zero-shot learning through cross-modal transfer. In Proc. NIPS2013. find
  47. R. Socher, A. Perelygin, J. Wu, J. Chuang, C. Manning, A. Ng, and C. Potts. 2013b. Recursive deep models for semantic compositionality over a sentiment treebank. In Proc. EMNLP2013. find
  48. Zhiyi Song, Ann Bies, Stephanie Strassel, Tom Riese, Justin Mott, Joe Ellis, Jonathan Wright, Seth Kulick, Neville Ryant, and Xiaoyi Ma. 2015. From light to rich ere: Annotation of entities, relations, and events. In Proc. NAACL-HLT2015 Workshop on on EVENTS. find our card
  49. P. Verga, D. Belanger, E. Strubell, B. Roth, and A. McCallum. 2016. Multilingual relation extrac-C. find
  50. Wang, N. Xue, and S. Pradhan. 2015a. Boosting transition-based amr parsing with refined actions and auxiliary analyzers. In Proc. ACL2015. find our card
  51. Chuan Wang, Nianwen Xue, Sameer Pradhan, and Sameer Pradhan. 2015b. A transition-based algorithm for amr parsing. In HLT-NAACL. find
  52. Yongqin Xian, Bernt Schiele, and Zeynep Akata. 2017. Zero-shot learning-the good, the bad and the ugly. arXiv preprint arXiv:1703.04394 . — [arXiv:1703.04394](https://arxiv.org/abs/1703.04394) find
  53. Z. Zhong and H. T. Ng. 2010. It makes sense: A wide-coverage word sense disambiguation system for free text. In Proc. ACL2010. find

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/huang-2018-zero-shot-transfer-learning-event/