基于机器阅读理解的事件检测方法

doi:10.13229/j.cnki.jdxbgxb.20220335

吉林大学学报(工学版) ›› 2024, Vol. 54 ›› Issue (2): 533-539.doi: 10.13229/j.cnki.jdxbgxb.20220335

• 计算机科学与技术 • 上一篇

基于机器阅读理解的事件检测方法

刘浏^1,²(),丁鲲¹(),刘姗姗¹,刘茗¹

^1.国防科技大学第六十三研究所，南京 210007
^2.宿迁学院信息工程学院，江苏宿迁 223800

收稿日期:2022-03-29 出版日期:2024-02-01 发布日期:2024-03-29
通讯作者: 丁鲲 E-mail:260344762@qq.com;dingkun18@nudt.edu.cn
作者简介:刘浏（1988-），男，讲师，博士. 研究方向：自然语言处理，人工智能，知识图谱.E-mail：260344762@qq.com
基金资助:
国家自然科学基金项目(71901215);江苏省“333工程”培养项目(BRA2020418);中国博士后科学基金项目(2021MD703983);江苏省高等学校自然科学研究面上项目(20KJB413003);宿迁市科技计划项目(K202128)

Event detection method as machine reading comprehension

Liu LIU^1,²(),Kun DING¹(),Shan-shan LIU¹,Ming LIU¹

^1.The Sixty-Third Research Institute，National University of Defense Technology，Nanjing 210007，China
^2.School of Information Engineering，Suqian University，Suqian 223800，China

Received:2022-03-29 Online:2024-02-01 Published:2024-03-29
Contact: Kun DING E-mail:260344762@qq.com;dingkun18@nudt.edu.cn

摘要/Abstract

摘要：

为提高事件检测任务的性能，将该任务重定义为一种提示范式，该范式使用问答对的形式将事件检测转化为机器阅读问题。同时，设计了一种名为WLBert-BiGRU的学习模型对问答对中的事件触发词进行预测，该模型使用Weight-Layers策略丰富Bert模型的语义表征能力，并使用双向门控循环单元神经网络（Bi-GRU）方法强化模型对事件触发词的识别能力。在ACE 2005数据集上的实验结果表明，本文方法在事件触发词识别和分类上的F₁指标分别达到了78.1%和75.1%，较现有的工作平均提高了4.18%和4.3%。

关键词: 人工智能, 事件检测, 自然语言处理, 机器阅读理解

Abstract:

In order to improve the performance of event detection task， this article redefines this task as a prompt paradigm. This paradigm uses question-answer pairs to transform event detection into machine reading problems. A pre-training model called WLBert-BiGRU is applied to predict the event triggers in QA pairs. The model uses Weight-Layers strategy to enrich the semantic representation ability of Bert model， and uses Bi-GRU to strengthen the prediction ability of the model to the event triggers. The proposed method is evaluated in ACE2005 data set， the results show that the F1 scores in event trigger recognition and classification have reached 78.1% and 75.1% respectively， which is 4.18% and 4.3% higher than the existing work.

Key words: artificial intelligence, event detection, natural language processing, machine reading comprehension

中图分类号:

TP391.1

刘浏,丁鲲,刘姗姗,刘茗. 基于机器阅读理解的事件检测方法[J]. 吉林大学学报(工学版), 2024, 54(2): 533-539.

Liu LIU,Kun DING,Shan-shan LIU,Ming LIU. Event detection method as machine reading comprehension[J]. Journal of Jilin University(Engineering and Technology Edition), 2024, 54(2): 533-539.

图/表 9

图1

图2

表1

图3

图4

表2

表3

表4

图5

参考文献 22

1	Linguistic Data Consortium. ACE (Automatic Content Extra-ction) English Annotation Guidelines for Events Version 5.4. 3 [EB/OL].[2005-07-01].
2	Du X, Cardie C. Event extraction by answering (almost) natural questions[C]∥Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing. Stroudsburg, PA: ACL, 2020: 671-683.
3	张婧丽. 基于知识扩展与信息融合的事件检测方法研究[D].苏州:苏州大学计算机科学与技术学院, 2020.
	Zhang Jing-li. Research on event detection based on knowledge expansion and information fusion[D]. Suzhou: School of Computer Science and Technology, Soochow University, 2020.
4	Grishman R, Westbrook D, Meyers A. Nyu's english ace 2005 system description[J]. Journal on Satisfiability, 2005, 51(11): 1927-1938.
5	Ahn D. The stages of event extraction[C]∥Proceedings of the Workshop on Annotating and Reasoning about Time and Events. Stroudsburg, PA: ACL, 2006: 1-8.
6	Ji H, Grishman R. Refining event extraction through cross-document inference[C]∥Proceedings of the 46th Annual Meeting of the Association for Computational Linguistics. Stroudsburg, PA: AcL, 2008: 254-262.
7	Liao S, Grishman R. Using document level cross-event inference to improve event extraction[C]∥Proceedings of the 48th Annual Meeting of the Association for Computational Linguistics. Stroudsburg, PA: AcL, 2010: 789-797.
8	Hong Y, Zhang J, Ma B, 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. Stroudsburg, PA: ACL, 2011: 1127-1136.
9	Li Q, Ji H, Huang L. Joint event extraction via structured prediction with global features[C]∥Proceedings of the 51st Annual Meeting of the Association for Computational Linguistics. Stroudsburg, PA: ACL, 2013: 73-82.
10	郭晓然, 罗平, 王维兰. 基于Transformer编码器的中文命名实体识别[J]. 吉林大学学报: 工学版, 2021, 51(3): 989-995.
	Guo Xiao-ran, Luo Ping, Wang Wei-lan. Chinese named entity recognition based on Transformer encoder[J]. Journal of Jilin University (Engineering and Technology Edition), 2021, 51(3): 989-995.
11	Nguyen T H, Grishman R. Modeling skip-grams for event detection with convolutional neural networks[C]∥Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing. Stroudsburg, PA: ACL, 2016: 886-891.
12	Chen Y, Xu L, Liu K, 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. Stroudsburg, PA: ACL, 2015: 167-176.
13	Nguyen T H, Cho K, Grishman R. 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. Stroudsburg, PA: ACL, 2016: 300-309.
14	Feng X, Qin B, Liu T. A language-independent neural network for event detection[J]. Science China Information Sciences, 2018, 61(9): 1-12.
15	Peters M E, Neumann M, Iyyer M, et al. Deep contextualized word representations[C]∥Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. Stroudsburg: ACL, PA, 2018: 2227-2237.
16	Radford A, Narasimhan K, Salimans T, et al. Improving language understanding by generative pre-training [EB/OL]. [2022-03-25].
17	Devlin J, Chang M W, Lee K, et al. Bert: Pre-training of deep bidirectional transformers for language understanding [EB/OL]. [2022-03-25].
18	李健, 熊琦, 胡雅婷, 等.基于Transformer和隐马尔科夫模型的中文命名实体识别方法[J/OL].[2022-05-12].DOI:10.13229/j.cnki.jdxbgxb20210856 . doi: 10.13229/j.cnki.jdxbgxb20210856
	Li Jian, Xiong Qi, Hu Ya-ting, et al. Chinese named entity recognition method based on transformer and hidden Markov model[J/OL]. [2022-05-12]. DOI:10.13229/j.cnki.jdxbgxb20210856 . doi: 10.13229/j.cnki.jdxbgxb20210856
19	Wadden D, Wennberg U, Luan Y, et al. Entity, relation, and event extraction with contextualized span representations[C]∥Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing. Stroudsburg, PA: ACL, 2019: 5784-5789.
20	Lin Y, Ji H, Huang F, et al. A joint neural model for information extraction with global features[C]∥Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics. Stroudsburg, PA: ACL, 2020: 7999-8009.
21	Nguyen T M, Nguyen T H. One for all: neural joint modeling of entities and events[C]∥Proceedings of the AAAI Conference on Artificial Intelligence, Menlo Park, CA, 2019: 6851-6858.
22	Zhang T, Ji H, Sil A. Joint entity and event extraction with generative adversarial imitation learning[J]. Data Intelligence, 2019, 1(2): 99-120.

相关文章 15

[1]	车翔玖,徐欢,潘明阳,刘全乐. 生物医学命名实体识别的两阶段学习算法[J]. 吉林大学学报(工学版), 2023, 53(8): 2380-2387.
[2]	李健,熊琦,胡雅婷,刘孔宇. 基于Transformer和隐马尔科夫模型的中文命名实体识别方法[J]. 吉林大学学报(工学版), 2023, 53(5): 1427-1434.
[3]	马月坤,郝益锋. 考虑特征稀疏特性的短文本命名实体快速识别方法[J]. 吉林大学学报(工学版), 2023, 53(12): 3529-3535.
[4]	刘春晖,王思长,郑策,陈秀连,郝春蕾. 基于深度学习的室内导航机器人避障规划算法[J]. 吉林大学学报(工学版), 2023, 53(12): 3558-3564.
[5]	白天,徐明蔚,刘思铭,张佶安,王喆. 基于深度神经网络的诉辩文本争议焦点识别[J]. 吉林大学学报(工学版), 2022, 52(8): 1872-1880.
[6]	王生生,姜林延,杨永波. 基于最优传输特征选择的医学图像分割迁移学习[J]. 吉林大学学报(工学版), 2022, 52(7): 1626-1638.
[7]	田皓宇,马昕,李贻斌. 基于骨架信息的异常步态识别方法[J]. 吉林大学学报(工学版), 2022, 52(4): 725-737.
[8]	刘勇,徐雷,张楚晗. 面向文本游戏的深度强化学习模型[J]. 吉林大学学报(工学版), 2022, 52(3): 666-674.
[9]	雷景佩,欧阳丹彤,张立明. 基于知识图谱嵌入的定义域值域约束补全方法[J]. 吉林大学学报(工学版), 2022, 52(1): 154-161.
[10]	李志华,张烨超,詹国华. 三维水声海底地形地貌实时拼接与可视化[J]. 吉林大学学报(工学版), 2022, 52(1): 180-186.
[11]	欧阳丹彤,张必歌,田乃予,张立明. 结合格局检测与局部搜索的故障数据缩减方法[J]. 吉林大学学报(工学版), 2021, 51(6): 2144-2153.
[12]	徐艳蕾,何润,翟钰婷,赵宾,李陈孝. 基于轻量卷积网络的田间自然环境杂草识别方法[J]. 吉林大学学报(工学版), 2021, 51(6): 2304-2312.
[13]	杨勇,陈强,曲福恒,刘俊杰,张磊. 基于模拟划分的SP⁃k⁃means-+算法[J]. 吉林大学学报(工学版), 2021, 51(5): 1808-1816.
[14]	赵亚慧,杨飞扬,张振国,崔荣一. 基于强化学习和注意力机制的朝鲜语文本结构发现[J]. 吉林大学学报(工学版), 2021, 51(4): 1387-1395.
[15]	董延华,刘靓葳,赵靖华,李亮,解方喜. 基于BPNN在线学习预测模型的扭矩实时跟踪控制[J]. 吉林大学学报(工学版), 2021, 51(4): 1405-1413.

Metrics

Viewed

Full text

Abstract

Cited

Shared

Discussed

问题模板	Which is the trigger in the sentence？
	What happened in the event？
	Which is the trigger？
	trigger
	verb

参数	数量（数值）
Epoch	6
Batch size	8
Learning rate	4×10^-5
Warm proportion	0.1
Dropout	0.1

模型	触发词识别			触发词分类
模型	p_i	r_i	f₁_i	p	r	f₁
DMCNN^［12］	80.4	67.7	73.5	75.6	63.6	69.1
JRNN^［13］	68.5	75.7	71.9	66.0	73.0	69.3
HNN^［14］	80.8	71.5	75.9	84.6	64.9	73.4
Joint3EE^［21］	70.5	74.5	72.5	68.0	71.8	69.8
baseline^［2］	74.3	77.4	75.8	71.1	73.7	72.4
WLBert-GRU	80.3	76	78.1	77.2	73.1	75.1

问题模板	触发词识别			触发词分类
问题模板	p_i	r_i	f₁_i	p	r	f₁
Which is the trigger in the sentence？	80.9	72.8	76.6	78.7	70.7	74.5
What happened in the event？	78.3	74.4	76.3	75.9	72.2	74.0
Which is the trigger？	76.3	76.0	76.2	73.9	73.6	73.7
trigger	74.6	77.8	76.2	72.5	75.6	74.0
verb	80.3	76.0	78.1	77.2	73.1	75.1
Bert+fine-tune	69.8	76.2	72.8	67.2	73.2	70.0

基于机器阅读理解的事件检测方法

Event detection method as machine reading comprehension

RICH HTML

PDF (PC)

摘要/Abstract

引用本文

使用本文

图/表 9

参考文献 22

相关文章 15

Metrics

本文评价

推荐阅读 0