Please wait a minute...
Advanced Search
数据分析与知识发现  2020, Vol. 4 Issue (9): 91-99     https://doi.org/10.11925/infotech.2096-3467.2020.0022
  研究论文 本期目录 | 过刊浏览 | 高级检索 |
扩充语义维度的BiGRU-AM突发事件要素识别研究*
尹浩然,曹金璇(),曹鲁喆,王国栋
中国人民公安大学信息网络安全学院 北京 100035
Identifying Emergency Elements Based on BiGRU-AM Model with Extended Semantic Dimension
Yin Haoran,Cao Jinxuan(),Cao Luzhe,Wang Guodong
College of Information Cyber Security, People’s Public Security University of China, Beijing 100035, China
全文: PDF (981 KB)   HTML ( 9
输出: BibTeX | EndNote (RIS)      
摘要 

目的】 为了解决循环神经网络对于重要程度不同的信息特征可解释性差的问题,本文提出一种扩充语义维度的BiGRU-AM突发事件要素识别方法。【方法】 首先将文本语料训练为词向量,并将生成的词向量联接词性、依存句法关系等语义特征;然后通过BiGRU模型提取上下文信息特征,将注意力机制引入BiGRU网络,使得特征的提取更有选择性;最后将学习到的特征经过Softmax函数激活,输出识别结果。【结果】 利用扩充了语义维度的BiGRU-AM模型在CEC数据集中进行实验,仿真实验结果表明,本文方法相较于其他浅层机器学习算法,F值提升了2%~21%不等。【局限】 在判断语义关系方面较为局限;语料预处理的过程中依赖分词工具的准确性;超参数只是单方面的依序确定,缺乏关联性;F值的提升使得预处理工作的开销增大。【结论】 扩充语义维度的BiGRU-AM模型可以有效地处理突发事件要素识别任务。

服务
把本文推荐给朋友
加入引用管理器
E-mail Alert
RSS
作者相关文章
尹浩然
曹金璇
曹鲁喆
王国栋
关键词 突发事件事件要素识别循环神经网络注意力机制    
Abstract

[Objective] This paper proposes a new method to recognize emergency elements based on a modified BiGRU-AM model, aiming to improve the poor interpretation of recurrent neural networks for information features with different degrees of importance.[Methods] First, we trained the text corpus to create word vectors, which were connected to semantic features like dependent syntactic relations. Then, we extracted contextual information features with BiGRU. We also introduced attention mechanism to the BiGRU network to extract diversified features. Finally, we activated the learned features with softmax function to generate needed elements.[Results] We examined the modified BiGRU-AM model with the CEC dataset and found its F-value was 2%-21% higher than algorithms of shallow machine learning.[Limitations] The proposed model’s ablilty to decide semantic relations, the accuracy of word segmentation tool, and the hyper parameters need to be improved.[Conclusions] The BiGRU-AM model with extended semantic dimension could effectively extract emergency elements.

Key wordsEmergency    Event Element Recognition    Recurrent Neural Network    Attention Mechanism
收稿日期: 2020-01-02      出版日期: 2020-07-17
ZTFLH:  TP183  
基金资助:*本文系中国人民公安大学基本科研业务费项目“人工智能在公安领域的应用”(2020JKF601);国家重点研发计划项目“社区风险监测与防范关键技术研究”的研究成果之一(2018YFC0809800)
通讯作者: 曹金璇     E-mail: caojinxuan@163.com
引用本文:   
尹浩然,曹金璇,曹鲁喆,王国栋. 扩充语义维度的BiGRU-AM突发事件要素识别研究*[J]. 数据分析与知识发现, 2020, 4(9): 91-99.
Yin Haoran,Cao Jinxuan,Cao Luzhe,Wang Guodong. Identifying Emergency Elements Based on BiGRU-AM Model with Extended Semantic Dimension. Data Analysis and Knowledge Discovery, 2020, 4(9): 91-99.
链接本文:  
http://manu44.magtech.com.cn/Jwk_infotech_wk3/CN/10.11925/infotech.2096-3467.2020.0022      或      http://manu44.magtech.com.cn/Jwk_infotech_wk3/CN/Y2020/V4/I9/91
Fig.1  事件要素识别模型
Fig.2  GRU的单元结构
事件类型 事件数量 句子数量 时间要素 地点要素 对象要素
地震 62 401 463 487 1 072
火灾 75 433 531 726 978
交通事故 85 514 697 1031 1 509
恐怖袭击 49 324 431 412 674
食物中毒 61 392 637 376 943
合计 332 2 064 2 759 3 032 5 176
Table 1  CEC语料库
维度 精确率/% 召回率/% F值/%
50 69.30 70.05 69.67
100 73.36 71.77 72.56
150 77.12 73.14 75.08
200 74.21 72.16 73.17
250 68.61 66.27 67.42
300 70.68 66.83 68.70
Table 2  不同词向量维度的识别效果
Fig.3  不同Dropout值的识别效果
Fig.4  不同Epoch值的识别效果
实验方法 精确率/% 召回率/% F值/%
POS[20] 68.80 47.70 56.34
触发词拓展[21] 65.28 63.49 64.37
BiGRU 74.59 71.20 72.86
依存句法分析[22] 72.67 73.71 73.19
MaxEnt+规则统计[23] 68.10 83.00 74.80
BiGRU-AM 76.29 73.63 74.93
要素填充+CRF[24] 79.40 73.70 76.00
ESD-BiGRU-AM 79.12 76.83 77.96
Table 3  实验对比结果
[1] 秦彦霞, 张民, 郑德权. 神经网络事件抽取技术综述[J]. 智能计算机与应用, 2018,8(3):1-5, 10.
[1] ( Qin Yanxia, Zhang Min, Zheng Dequan. A Survey on Neural Network-Based Methods for Event Extraction[J]. Intelligent Computer and Applications, 2018,8(3):1-5, 10.)
[2] 夏彦, 何琳, 潘运来, 等. 基于规则与统计相结合的互联网突发事件识别研究[J]. 现代图书情报技术, 2010(10):65-69.
[2] ( Xia Yan, He Lin, Pan Yunlai, et al. Research on Recognition of Sudden Events on Web Based on Combination of Rules and Statistical Method[J]. New Technology of Library and Information Service, 2010(10):65-69.)
[3] Zhong Z Y, Jin L W, Feng Z Y. Multi-font Printed Chinese Character Recognition Using Multi-pooling Convolutional Neural Network[C] //Proceedings of the 13th International Conference on Document Analysis and Recognition. DOI: 10.1109/ICDAR.2015.7333733.
[4] 蒋德良. 基于规则匹配的突发事件结果信息抽取研究[J]. 计算机工程与设计, 2010,31(14):3294-3297.
[4] ( Jiang Deliang. Research on Extraction of Emergency Event Information Based on Rules Matching[J]. Computer Engineering and Design, 2010,31(14):3294-3297.)
[5] 姚占雷, 许鑫. 互联网新闻报道中的突发事件识别研究[J]. 现代图书情报技术, 2011(4):52-57.
[5] ( Yao Zhanlei, Xu Xin. Research on the Detection of Sudden Events in News Stories of Online Information[J]. New Technology of Library and Information Service, 2011(4):52-57.)
[6] 何中市, 刘莉, 邢欣来, 等. 基于语义角色的中文事件识别[J]. 计算机工程与科学, 2013,35(4):181-185.
[6] ( He Zhongshi, Liu Li, Xing Xinlai, et al. Chinese Events Recognition Based on Semantic Role[J]. Computer Engineering & Science, 2013,35(4):181-185.)
[7] 唐慧慧, 王昊, 张紫玄, 等. 基于汉字标注的中文历史事件名抽取研究[J]. 数据分析与知识发现, 2018,2(7):89-100.
[7] ( Tang Huihui, Wang Hao, Zhang Zixuan, et al. Extracting Names of Historical Events Based on Chinese Character Tags[J]. Data Analysis and Knowledge Discovery, 2018,2(7):89-100.)
[8] 黎红. 基于深度学习的维吾尔语文本事件信息的抽取与分析[D]. 乌鲁木齐: 新疆大学, 2018.
[8] ( Li Hong. Extraction and Analysis of Event Information in Uyghur Language Based on Deep Learning[D]. Urumqi: Xinjiang University, 2018.)
[9] 施喆尔, 陈锦秀. 基于语言模型及循环卷积神经网络的事件检测[J]. 厦门大学学报(自然科学版), 2019,58(3):442-448.
[9] ( Shi Zheer, Chen Jinxiu. Event Detection via Recurrent and Convolutional Networks Based on Language Model[J]. Journal of Xiamen University (Natural Science), 2019,58(3):442-448.)
[10] Liu K, Sun Z S, Jiang J F, et al. A Combined Events Recognition Scheme Using Hybrid Features in Distributed Optical Fiber Vibration Sensing System[J]. IEEE Access, 2019. DOI: 10.1109/ACCESS.2019.2932187.
doi: 10.1109/ACCESS.2019.2958985 pmid: 31942279
[11] 黄细凤. 基于动态掩蔽注意力机制的事件抽取[J/OL]. 计算机应用研究. [2019-07-27]. https://doi.org/10.19734/j.issn.1001-3695.
[11] ( Huang Xifeng. Event Extraction via Dynamic Masked Attention[J/OL]. Application Research of Computers. [2019-07-27]. https://doi.org/10.19734/j.issn.1001-3695.2018.12.0927.)
[12] 张亚军, 刘宗田, 周文. 基于深度信念网络的事件识别[J]. 电子学报, 2017,45(6):1415-1423.
[12] ( Zhang Yajun, Liu Zongtian, Zhou Wen. Event Recognition Based on Deep Belief Network[J]. Acta Electronica Sinica, 2017,45(6):1415-1423.)
[13] Lin H Y, Lu Y J, Han X P, et al. Nugget Proposal Networks for Chinese Event Detection[OL]. arXiv Preprint, arXiv: 1805.00249.
[14] 涂铭, 刘祥, 刘树春. Python自然语言处理实战: 核心技术与算法[M]. 北京: 机械工业出版社, 2018.
[14] ( Tu Ming, Liu Xiang, Liu Shuchun. Natural Language Processing: Core Technology and Algorithm with Python[M]. Beijing: China Machine Press, 2018.)
[15] Mikolov T, Sutskever I, Chen K, et al. Distributed Representations of Words and Phrases and Their Compositionality[C] //Proceedings of the 26th International Conference on Neural Information Processing Systems. 2013: 3111-3119.
[16] Mikolov T, Chen K, Corrado G, et al. Efficient Estimation of Word Representations in Vector Space[OL]. arXiv Preprint, arXiv: 1301.3781.
[17] Cho K, van Merrienboer B, Gulcehre C, et al. Learning Phrase Representations Using RNN Encoder-Decoder for Statistical Machine Translation[OL]. arXiv Preprint, arXiv: 1406.1078.
[18] Mnih V, Heess N, Graves A, et al. Recurrent Models of Visual Attention[C] //Proceedings of 2014 Advances in Neural Information Processing Systems. 2014: 2204-2212.
[19] Bahdanau D, Cho K, Bengio Y. Neural Machine Translation by Jointly Learning to Align and Translate[C] //Proceedings of International Conference on Learning Representations, 2015.
[20] Ahn D. The Stages of Event Extraction[C] //Proceedings of the Workshop on Annotating and Reasoning about Time and Events. 2006: 1-8.
[21] 赵妍妍, 秦兵, 车万翔, 等. 中文事件抽取技术研究[J]. 中文信息学报, 2008,22(1):3-8.
[21] ( Zhao Yanyan, Qin Bing, Che Wanxiang, et al. Research on Chinese Event Extraction[J]. Journal of Chinese Information Processing, 2008,22(1):3-8.)
[22] 张鹏. 面向突发事件的事件要素识别研究[D]. 淮南:安徽理工大学, 2018.
[22] ( Zhang Peng. Research on Event Element Recognition for Emergencies[D]. Huainan: Anhui University of Science & Technology, 2018.)
[23] 孙佩佩. 面向事件的自动文摘研究[D]. 淮南: 安徽理工大学, 2016.
[23] ( Sun Peipei. Research on Event-oriented Automatic Summarization[D]. Huainan: Anhui University of Science & Technology, 2016.)
[24] 张璐. 面向中文文本的事件提取方法研究[D]. 北京: 中国人民公安大学, 2019.
[24] ( Zhang Lu. Research on Chinese-oriented Text Event Extraction[D]. Beijing: People’s Public Security University of China, 2019.)
[25] 邱奇志, 周三三, 刘长发, 等. 基于文体和词表的突发事件信息抽取研究[J]. 中文信息学报, 2018,32(9):56-65, 74.
[25] ( Qiu Qizhi Zhou Sansan Liu Changfa, et al. Emergency Information Extraction Based on Style and Terminology[J]. Journal of Chinese Information Processing, 2018,32(9):56-65, 74.)
[26] Yin H R, Cao J X, Cao L Z, et al. Chinese Emergency Event Recognition Using Conv-RDBiGRU Model[J]. Computational Intelligence and Neuroscience, 2020: 7090918.
[27] 闻畅, 刘宇, 顾进广. 基于注意力机制的双向长短时记忆网络模型突发事件演化关系抽取[J]. 计算机应用, 2019,39(6):1646-1651.
[27] ( Wen Chang, Liu Yu, Gu Jinguang. Evolution Relationship Extraction of Emergency Based on Attention-based Bidirectional Long Short-term Memory Network Model[J]. Journal of Computer Applications, 2019,39(6):1646-1651.)
[28] Li D Y, Huang L F, Ji H, et al. Biomedical Event Extraction Based on Knowledge-driven Tree-LSTM[C] // Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers). 2019: 1421-1430.
[29] Chen Y, Zhou C J, Li T X, et al. Named Entity Recognition from Chinese Adverse Drug Event Reports with Lexical Feature Based BiLSTM-CRF and Tri-training[J]. Journal of Biomedical Informatics, 2019,96:103252.
doi: 10.1016/j.jbi.2019.103252 pmid: 31323311
[1] 黄露,周恩国,李岱峰. 融合特定任务信息注意力机制的文本表示学习模型*[J]. 数据分析与知识发现, 2020, 4(9): 111-122.
[2] 石磊,王毅,成颖,魏瑞斌. 自然语言处理中的注意力机制研究综述*[J]. 数据分析与知识发现, 2020, 4(5): 1-14.
[3] 苏传东,黄孝喜,王荣波,谌志群,毛君钰,朱嘉莹,潘宇豪. 基于词嵌入融合和循环神经网络的中英文隐喻识别*[J]. 数据分析与知识发现, 2020, 4(4): 91-99.
[4] 邓建高,张璇,傅柱,韦庆明. 基于系统动力学的突发事件网络舆情传播研究:以“江苏响水爆炸事故”为例*[J]. 数据分析与知识发现, 2020, 4(2/3): 110-121.
[5] 梁艳平,安璐,刘静. 同类突发公共卫生事件微博话题共振研究*[J]. 数据分析与知识发现, 2020, 4(2/3): 122-133.
[6] 薛福亮,刘丽芳. 一种基于CRF与ATAE-LSTM的细粒度情感分析方法*[J]. 数据分析与知识发现, 2020, 4(2/3): 207-213.
[7] 吴粤敏,丁港归,胡滨. 基于注意力机制的农业金融文本关系抽取研究*[J]. 数据分析与知识发现, 2019, 3(5): 86-92.
[8] 胡哲,查先进,严亚兰. 突发事件情境下在线健康社区用户交互行为研究 *[J]. 数据分析与知识发现, 2019, 3(12): 10-20.
[9] 陶志勇,李小兵,刘影,刘晓芳. 基于双向长短时记忆网络的改进注意力短文本分类方法 *[J]. 数据分析与知识发现, 2019, 3(12): 21-29.
[10] 李纲,陈思菁,毛进,谷岩松. 自然灾害事件微博热点话题的时空对比分析 *[J]. 数据分析与知识发现, 2019, 3(11): 1-15.
[11] 李钰曼,陈志泊,许福. 基于KACC模型的文本分类研究 *[J]. 数据分析与知识发现, 2019, 3(10): 89-97.
[12] 丁晟春,龚思兰,李红梅. 基于突发主题词和凝聚式层次聚类的微博突发事件检测研究*[J]. 现代图书情报技术, 2016, 32(7-8): 12-20.
[13] 吴鹏,金贝贝,强韶华. 基于BDI-Agent模型的突发事件网络舆情应急响应建模研究*[J]. 现代图书情报技术, 2016, 32(7-8): 32-41.
[14] 朱丹浩, 杨蕾, 王东波. 基于深度学习的中文机构名识别研究*——一种汉字级别的循环神经网络方法[J]. 数据分析与知识发现, 2016, 32(12): 36-43.
[15] 吴鹏, 杨爽, 张晶晶, 高庆宁. 突发事件网络舆情中网民群体行为演化的Agent建模与仿真研究[J]. 现代图书情报技术, 2015, 31(7-8): 65-72.
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
版权所有 © 2015 《数据分析与知识发现》编辑部
地址:北京市海淀区中关村北四环西路33号 邮编:100190
电话/传真:(010)82626611-6626,82624938
E-mail:jishu@mail.las.ac.cn