句子级事件检测任务目的是识别和分类事件触发词。现阶段工作主要将句子作为神经分类网络的输入,学习句子的深层语义信息,从而优化句子表示来改进事件检测任务的性能。该文发现除句子语义信息外,依存树包含的句法结构信息也有助于获取准确的句子表示。为此,该文采用双向长短时记忆网络对句子进行编码,捕获其语义信息;同时,设计图神经网络对句子的依存结构进行表示,获取其依存信息;此外,在对句子进行语义编码与依存编码时,该文利用自注意力机制使模型选择性地关注句子中的不同词,从而捕获句子中有助于事件检测的关键信息,并尽可能避免无关词的干扰;最后,该文提出门控机制,通过加权实现上述两种信息的动态融合。该文在自动文本抽取(automatic content extraction, ACE)数据集上进行实验,结果显示,该文提出的动态融合语义信息与依存信息的方法能更加有效地对句子进行编码,并捕获句子中的事件信息,在触发词识别与事件类型分类这两个子任务中,F1值均有较大提升,分别达到76.3%和73.9%。
Abstract
Sentence-level Event Detection (ED) is a task of identifying and classifying event triggers. Existing approaches mainly use sentences as the input of the neural classification network and learn the deep semantic information of sentences. Base on the fact that the dependency tree contains rich syntactic structure features for more accurate sentence representation, we use a Bidirectional Long Short-Term Memory (Bi-LSTM) to learn semantic information, and use a Graph Convolutional Network (GCN) to learn dependency information. To concentrate more on event-related information and reduce the interference of redundant words, we add self-attention on the Bi-LSTM and GCN respectively. Finally, we propose to use the gated mechanism to dynamically fuse semantic information and dependency information. The experimental results on ACE show that the performance of the proposed method reaches 76.3% and 73.9% in F1-score for trigger identification and event type classification, respectively.
关键词
语义信息 /
依存信息 /
门控机制 /
事件检测
{{custom_keyword}} /
Key words
semantic information /
dependency information /
gated mechanism /
event detection
{{custom_keyword}} /
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
[1] Nguyen Thien Huu, Cho Kyunghyun, Grishman Ralph. Joint event extraction via recurrent neural networks[C]//Proceedings of the 2016 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, 2016: 300-309.
[2] Liu Xiao, Luo Zhunchen, Huang Heyan. Jointly multiple events extraction via attention-based graph information aggregation[J]. arXiv preprint arXiv: 1809.09078, 2018.
[3] Schuster Mike,Paliwal Kuldip K. Bidirectional recurrent neural networks[J]. IEEE Transactions on Signal Processing, 1997, 45(11): 2673-2681.
[4] Kipf Thomas N, Max Welling. Semi-supervised classification with graph convolutional networks[J]. arXiv preprint arXiv: 1609.02907, 2016.
[5] Vaswani Ashish, Shazeer Noam, Parmar Niki, et al. Attention is all you need[J].arXiv preprint arXiv:1706.03762, 2017.
[6] Grishman Ralph, Westbrook David, Meyers Adam. NYUs English ACE 2005 system description[J]. ACE, 2005, 5.
[7] Ahn David. The stages of event extraction[C]//Proceedings of the Workshop on Annotating and Reasoning about Time and Events, 2006: 1-8.
[8] Ji Heng, Grishman Ralph. Refining event extraction through cross-document inference[C]//Proceedings of ACL-08: HLT, 2008: 254-262.
[9] Gupta Prashant,Ji Heng. Predicting unknown time arguments based on cross-event propagation[C]//Proceedings of the ACL-IJCNLP 2009 Conference Short Papers, 2009: 369-372.
[10] Grishman Ralph. The impact of task and corpus on event extraction systems[C]//Proceedings of LREC, 2010.
[11] Hong Yu, Zhang Jianfeng, Ma Bin, et al. Using cross-entity inference to improve event extraction[C]//Proceedings of the 49th Annual Meeting of the Association for Computational Linguistics: Human Language Technologies. Association for Computational Linguistics, 2011: 1127-1136.
[12] Liu Shulin, Chen Yubo, He Shizhu, et al. Leveraging framenet to improve automatic event detection[C]//Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics, 2016, 1: 2134-2143.
[13] Zeng Ying, Feng Yansong, Ma Rong, et al. Scale up event extraction learning via automatic training data generation[C]//Proceedings of the 32 AAAI Conference on Artificial Intelligence, 2018.
[14] Nguyen Thien Huu, Grishman Ralph. Event detection and domain adaptation with convolutional neural networks[C]//Proceedings of the 53rd Annual Meeting of the Association for Computational Linguistics and the 7th International Joint Conference on Natural Language Processing, 2015, 2: 365-371.
[15] Chen Yubo, Xu Liheng, Liu Kang, et al. Event extraction via dynamic multi-pooling convolutional neural networks[C]//Proceedings of the 53rd Annual Meeting of the Association for Computational Linguistics and the 7th International Joint Conference on Natural Language Processing, 2015, 1: 167-176.
[16] Nguyen Thien Huu, Grishman Ralph. Modeling skip-grams for event detection with convolutional neural networks[C]//Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing, 2016: 886-891.
[17] Duan Shaoyang, He Ruifang, Zhao Wenli. Exploiting document level information to improve event detection via recurrent neural networks[C]//Proceedings of the 8th International Joint Conference on Natural Language Processing, 2017, 1: 352-361.
[18] Feng Xiaocheng, Qin Bing, Liu Ting. A language-independent neural network for event detection[J]. Science China Information Sciences, 2018, 61(9): 092-106.
[19] Chen Yubo, Yang Hang, Liu Kang, et al. Collective event detection via a hierarchical and bias tagging networks with gated multi-level attention mechanisms[C]//Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing, 2018: 1267-1276.
[20] Zhao Yue, Jin Xiaolong, Wang Yuanzhuo, et al. Document embedding enhanced event detection with hierarchical and supervised attention[C]//Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics, 2018, 2: 414-419.
[21] Hong Yu, Zhou Wenxuan, Zhang Jingli, et al. Self-regulation: Employing a generative adversarial network to improve event detection[C]//Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics, 2018, 1: 515-526.
[22] Liu Jian, ChenYubo, Liu Kang, et al. Event detection via gated multilingual attention mechanism[C]//Proceedings of the 32nd AAAI Conference on Artificial Intelligence, 2018.
[23] Li Qi,Ji Heng, Huang Liang. Joint event extraction via structured prediction with global features[C]//Proceedings of the 51st Annual Meeting of the Association for Computational Linguistics, 2013, 1: 73-82.
[24] Sha Lei, Qian Feng, Chang Baobao, et al. Jointly extracting event triggers and arguments by dependency-bridge RNN and tensor-based argument interaction[C]//Proceedings of 32nd AAAI Conference on Artificial Intelligence, 2018.
[25] Cao Yu, Meng Fang, Tao Dacheng. BAG: Bi-directional attention entity graph convolutional network for multi-hop reasoning question answering[J].arXiv preprint arXiv: 1904.04969, 2019.
[26] Guo Zhijiang, Zhang Yan, Lu Wei. Attention guided graph convolutional networks for relation extraction[J]. arXiv preprint arXiv: 1906.07510, 2019.
[27] Doddington George R, Mitchell Alexis, Przybocki Mark A, et al. The automatic content extraction (ACE) program-tasks, data, and evaluation[C]//Proceedings of the 4th International Conference on Langnage Resources and Evahuation, 2004, 2: 1.
[28] Mikolov Tomas, Chen Kai, Corrado Greg, et al. Efficient estimation of word representations in vector space[J]. arXiv preprint arXiv: 1301.3781, 2013.
[29] Hochreiter Sepp, Schmidhuber Jürgen. Long short-term memory[J]. Neural Computation, 1997, 9(8): 1735-1780.
[30] Liao Shasha, Grishman Ralph. Using document level cross-event inference to improve event extraction[C]//Proceedings of the 48th Annual Meeting of the Association for Computational Linguistics. Association for Computational Linguistics, 2010: 789-797.
[31] Liu Shulin, Liu Kang, He Shizhu, et al. A probabilistic soft logic based approach to exploiting latent and global information in event classification[C]//Proceedings of the 30th AAAI Conference on Artificial Intelligence, 2016.
[32] Liu Shulin, Chen Yubo, Liu Kang, et al. Exploiting argument information to improve event detection via supervised attention mechanisms[C]//Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics, 2017:1789-1798.
{{custom_fnGroup.title_cn}}
脚注
{{custom_fn.content}}
基金
国家自然科学基金(61672368,61672367,61703293)
{{custom_fund}}
PDF(1213 KB)
