事件共指关系识别旨在分析事件描述之间是否从不同的角度对同一件真实事件展开论述。但是,在同一篇新闻报道中往往存在不同事件句之间具有相似上下文但不具有共指关系的噪声情况,其会对共指关系识别模型造成干扰。为解决以上问题,该文提出了基于生成对抗网络的越南语新闻事件共指关系识别方法,采用触发词的上下文信息作为事件句的最小特征表示,在生成对抗网络的基础上构建噪声数据过滤机制进行信息实例与噪声实例的区分。在越南语事件数据集和公开数据集上的实验表明,该神经网络模型能有效进行噪声数据过滤,相对于传统的事件共指关系识别方法有明显的优势。
Abstract
Event coreference resolution is a task to analyze whether event descriptions discuss the same real event from different perspectives. However, in the same news report, there is usually a noisy situation in which different event sentences have similar contexts but no coreference relationship. To address this issue, we propose an event coreference resolution method based on generative adversarial networks. It uses the context information of trigger words as the minimum feature representation of event sentence, and adopts the generative adversarial network to construct noise data filters to distinguish information instances from noise instances. Experiments on the Vietnamese event dataset and public dataset verify the validity of the model.
关键词
越南语新闻 /
事件共指关系识别 /
生成对抗网络
{{custom_keyword}} /
Key words
Vietnamese news /
event coreference resolution /
generative adversarial networks
{{custom_keyword}} /
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
[1] ZUO X Y, CHEN Y B, LIU K, et al. Event co-reference resolution via a multi-loss neural network without using argument information[J].Science China Information Sciences, 2019, 62(11): 1-9.
[2] PANDIAN A, MULAFFER L, OFLAZER K, et al. Event coreference resolution using neural network classifiers[J]. arXiv preprint arXiv: 1810.04216, 2018.
[3] FANG J, LI P F, ZHOU G D. Employing multiple decomposable attention networks to resolve event coreference[C]//Proceedings of CCF International Conference on Natural Language Processing and Chinese Computing. 2018: 246-256.
[4] 陆震寰, 孔芳, 周国栋. 面向多语料库的通用事件指代消解[J].中文信息学报, 2018, 32(1): 51-58.
[5] GOODFELLOW I, POUGET ABADIE J, MIRZA M,et al. Generative adversarial nets[C]//Proceedings of the 27th International Conference on Neural Information Processing Systems. Montreal, Canada: MIT Press, 2014: 2672-2680.
[6] HONG Y, ZHOU W X, ZHANG J L,et al. Self-regulation: Employing a generative advers arial network to improve event detection[C]//Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics, 2018: 515-526.
[7] ZHANG T T, JI H, SIL A. Joint Entity and event extraction with generative adversarial imitation learning[J]. Data Intell.2019, 1(2): 99-120.
[8] WANG X Z, HAN X, LIU Z Y, et al. Adversarial training for weakly supervised event detection[C]//Proceedings of the Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, 2019: 998-1008.
[9] XU H, LIU Z, SUN M. Denoising distant supervision for relation extraction via instance-level adversarial training[J]. arXiv preprint arXiv: 1805.10959, 2018.
[10] CHEN B, JIAN S, PAN S J, et al. A unified event coreference resolution by integrating multiple resolvers[C]//Proceedings of International Joint Conference on Natural Language Processing. Chiang Mai, 2011: 102-110.
[11] LU J, NG V. Joint learning for event coreference resolution[C]//Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics. 2017: 90-101.
[12] LU J, NG V. Learning antecedent structures for event coreference resolution[C]//Proceedings of the 16th IEEE International Conference on Machine Learning and Applications. IEEE, 2017: 113-118.
[13] LIU Z, ARAKI J, HOVY E, et al. Supervised within-document event coreference using information propagation[C]//Proceedings of LREC, 2014: 4539-4544.
[14] CYBULSKA A, VOSSEN P. Translating Granularity of event slots into features for event coreference resolution[C]//Proceedings of the the 3rd Workshop on EVENTS: Definition,Detection, Coreference and Representation. 2015: 1-10.
[15] AHN D.The stages of event extraction[C]//Proceedings of the Workshop on Annotating and Reasoning about Time and Events. 2006: 1-8.
[16] CHEN C. Sinocoreferencer: An end-to-end Chinese event coreference resolver[C]//Proceedings of the Ninth Language Resources and Evaluation Conference, 2014: 4532-4538.
[17] NGUYEN D Q, NGUYEN A T. PhoBERT: Pre-trained language models for Vietnamese[C]//Proceedings of the Conference on Empirical Methods in Natural Language Processing: Findings, 2020: 1037-1042.
[18] CYBULSKA A, VOSSEN P. Guidelines for ECB+ annotation of events and their coreference[J]. Technical Report,2014: 1-28.
[19] LAI T, JI H, BUI T, et al. A context-dependent gated module for incorporating symbolic semantics into event coreference resolution[C]//Proceedings of the Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies,2021: 3491-3499.
[20] AHMED S R, MARTIN J H.Within-document event coreference with BERT-based contextualized representations[J]. arXiv preprint arXiv: 2102.09600, 2021.
{{custom_fnGroup.title_cn}}
脚注
{{custom_fn.content}}
基金
国家自然科学基金(61972186,U21B2027,61732005);云南省重大科技专项(202002AD080001,202202AD080003,202103AA080015);云南省高新技术产业专项(201606)
{{custom_fund}}
PDF(2140 KB)
