面向行政执法案件文本的事件抽取研究

doi:10.19678/j.issn.1000-3428.0068523

摘要/Abstract

摘要：

行政执法的智能化水平是国家治理能力现代化的体现, 数据是智能化发展的重要依托。在行政执法领域, 各行政机关存储大量以文本形式记录的历史案件, 这种非结构化的数据价值密度较低、可利用性不强。利用事件抽取技术从行政执法案件文本中快速高效地抽取案件职权类型、案发时间、案发地点等结构化信息, 可推动行政机关对历史案件信息的利用和智能化执法办案研究。收集整理某城市的真实案例数据, 并通过人工标注构建一个行政执法领域的数据集, 根据行政执法案件文本的无触发词、文档级、格式不固定等文本特征, 提出结合基于Transformer的双向编码器表示(BERT)和基于条件随机场的双向长短期记忆网络(BiLSTM-CRF)模型的两阶段事件抽取方法, 通过文本多分类和序列标注依次完成事件类型检测和事件论元抽取任务。实验结果表明, 事件类型检测任务的F1值达到99.54%, 事件论元抽取任务的F1值达到97.36%, 实现了对案件信息的有效抽取。

关键词: 行政执法案件, 事件抽取, 两阶段方法, 基于Transformer的双向编码器表示模型, 基于条件随机场的双向长短期记忆网络(BiLSTM-CRF)模型

Abstract:

The level of intelligence in administrative law enforcement is a manifestation of the modernization of national governance capacity, and data is an important support for the development of intelligence. In the field of administrative law enforcement, various administrative organs store numerous historical cases recorded in textual form. These cases are unstructured data with low value density and limited usability. The use of event extraction technology for the quick and efficient extraction of structured information, such as the type of case authority and the time and place of case occurrence, from administrative law enforcement case texts can promote the utilization of historical case records and provide support for the study of intelligent law enforcement. This study collects and organizes real case data for a city and constructs a dataset in the field of administrative law enforcement through manual annotation. Considering text characteristics, such as no trigger words, document-level text, and unfixed format, the study then proposes a two-stage event extraction method based on a Bidirectional Encoder Representations from Transformers (BERT) model and a Bi-directional Long Short-Term Memory network with Conditional Random Field (BiLSTM-CRF) model, which sequentially detects event types and identifies event arguments through text multi-classification and sequence annotation. Experimental results show that the F1 values of event-type detection and event-argument extraction tasks reach 99.54% and 97.36%, respectively, thus realizing the effective extraction of case information.

Key words: administrative law enforcement case, event extraction, two-stage method, Bidirectional Encoder Representations from Transformers (BERT) model, Bi-directional Long Short-Term Memory network with Conditional Random Field (BiLSTM-CRF) model

屈潇雅, 李兵, 温立强. 面向行政执法案件文本的事件抽取研究[J]. 计算机工程, 2024, 50(9): 63-71.

QU Xiaoya, LI Bing, WEN Liqiang. Research on Event Extraction for Administrative Law Enforcement Case Texts[J]. Computer Engineering, 2024, 50(9): 63-71.

https://www.ecice06.com/CN/Y2024/V50/I9/63

图/表 14

图1 行政执法案件文本示例

Fig.1 An example of administrative law enforcement case text

图2 文本长度分布图

Fig.2 Distribution map of text length

图3 事件抽取模型框架

Fig.3 Framework of event extraction model

图4 BERT模型框架

Fig.4 Framework of BERT model

图5 BiLSTM模型结构

Fig.5 Structure of BiLSTM model

图6 BERT模型混淆矩阵

Fig.6 Confusion matrix of BERT model

图7 事件类型检测错误示例

Fig.7 Examples of detecting incorrect event types

参考文献 26

1	LI Q, LI J X, SHENG J W, et al. A survey on deep learning event extraction: approaches and applications. IEEE Transactions on Neural Networks and Learning Systems, 2024, 35(5): 6301- 6321. doi: 10.1109/TNNLS.2022.3213168
2	马春明, 李秀红, 李哲, 等. 事件抽取综述. 计算机应用, 2022, 42(10): 2975- 2989. URL
	MA C M, LI X H, LI Z, et al. Survey of event extraction. Journal of Computer Applications, 2022, 42(10): 2975- 2989. URL
3	RILOFF E. Automatically constructing a dictionary for information extraction tasks[C]//Proceedings of the 11th International Conference on Artificial Intelligence. Palo Alto, USA: AAAI Press, 1993: 811-816.
4	陈诺, 李旭晖. 一种基于模板提示学习的事件抽取方法. 数据分析与知识发现, 2023, 7(6): 86- 98. URL
	CHEN N, LI X H. An event extraction method based on template prompt learning. Data Analysis and Knowledge Discovery, 2023, 7(6): 86- 98. URL
5	陈清政. 基于编码解析与模式匹配的法院案件信息抽取研究[D]. 南昌: 江西财经大学, 2023.
	CHEN Q Z. Research on information extraction of court cases based on code analysis and pattern matching[D]. Nanchang: Jiangxi University of Finance and Economics, 2023. (in Chinese)
6	CHIEU H L, NG H T. A maximum entropy approach to information extraction from semi-structured and free text[C]//Proceedings of the 18th International Conference on Artificial Intelligence. New York, USA: ACM Press, 2002: 786-791.
7	LYBARGER K, OSTENDORF M, THOMPSON M, et al. Extracting COVID-19 diagnoses and symptoms from clinical text: a new annotated corpus and neural event extraction framework. Journal of Biomedical Informatics, 2021, 117, 103761. doi: 10.1016/j.jbi.2021.103761
8	CHEN Y B, XU L H, 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 (Volume 1: Long Papers). Stroudsburg, USA: Association for Computational Linguistics, 2015: 167-176.
9	FENG X C, QIN B, LIU T. A language-independent neural network for event detection. Science China Information Sciences, 2018, 61(9): 092106. doi: 10.1007/s11432-017-9359-x
10	陈斌, 周勇, 刘兵. 基于卷积双向长短期记忆网络的事件触发词抽取. 计算机工程, 2019, 45(1): 153- 158. URL
	CHEN B, ZHOU Y, LIU B. Event trigger word extraction based on convolutional bidirectional long short term memory network. Computer Engineering, 2019, 45(1): 153- 158. URL
11	WEI H, ZHOU A, ZHANG Y J, et al. Biomedical event trigger extraction based on multi-layer residual BiLSTM and contextualized word representations. International Journal of Machine Learning and Cybernetics, 2022, 13(3): 721- 733. doi: 10.1007/s13042-021-01315-7
12	LIU S, LI Y, ZHANG F, et al. Event detection without triggers[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). Stroudsburg, USA: Association for Computational Linguistics, 2019: 735-744.
13	YANG S, FENG D W, QIAO L B, et al. Exploring pre-trained language models for event extraction and generation[C]//Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics. Stroudsburg, USA: Association for Computational Linguistics, 2019: 5284-5294.
14	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, USA: Association for Computational Linguistics, 2016: 300-309.
15	ZHENG S, CAO W, XU W, et al. Doc2EDAG: an end-to-end document-level framework for Chinese financial event extraction[EB/OL]. [2023-09-15]. http://arxiv.org/abs/1904.07535.
16	王雷, 李瑞轩, 李玉华, 等. 文档级无触发词事件抽取联合模型. 计算机科学与探索, 2021, 15(12): 2327- 2334. URL
	WANG L, LI R X, LI Y H, et al. Joint model for document-level event extraction without triggers. Journal of Frontiers of Computer Science and Technology, 2021, 15(12): 2327- 2334. URL
17	VASWANI A, SHAZEER N, PARMAR N, et al. Attention is all you need[EB/OL]. [2023-09-15]. http://arxiv.org/abs/1706.03762.
18	XU R X, LIU T Y, LI L, et al. Document-level event extraction via heterogeneous graph-based interaction model with a tracker[EB/OL]. [2023-09-15]. http://arxiv.org/abs/2105.14924.
19	周晓进, 徐陈铭, 阮彤. 面向中文电子病历的多粒度医疗实体识别. 计算机科学, 2021, 48(4): 237- 242. URL
	ZHOU X J, XU C M, RUAN T. Multi-granularity medical entity recognition for Chinese electronic medical records. Computer Science, 2021, 48(4): 237- 242. URL
20	王剑, 彭雨琦, 赵宇斐, 等. 基于深度学习的社交网络舆情信息抽取方法综述. 计算机科学, 2022, 49(8): 279- 293. URL
	WANG J, PENG Y Q, ZHAO Y F, et al. Survey of social network public opinion information extraction based on deep learning. Computer Science, 2022, 49(8): 279- 293. URL
21	胡婕, 何巍, 曾张帆. 基于RoBERTa的全局图神经网络文档级中文金融事件抽取. 中文信息学报, 2023, 37(2): 107- 118. URL
	HU J, HE W, ZENG Z F. Document-level Chinese financial event extraction based on RoBERTa and global graph neural network. Journal of Chinese Information Processing, 2023, 37(2): 107- 118. URL
22	张琳, 秦策, 叶文豪. 基于条件随机场的法言法语实体自动识别模型研究. 数据分析与知识发现, 2017,(11): 46- 52. URL
	ZHANG L, QIN C, YE W H. Automatic recognition of legal language entities based on conditional random fields. Data Analysis and Knowledge Discovery, 2017,(11): 46- 52. URL
23	汪瀛寰. 面向司法判决书案情的主题事件抽取研究[D]. 合肥: 合肥工业大学, 2019.
	WANG Y H. Research on topic event extraction for court verdict case[D]. Hefei: Hefei University of Technology, 2019. (in Chinese)
24	贾阵, 丁泽华, 陈艳平, 等. 面向司法数据的事件抽取方法研究. 计算机工程与应用, 2023, 59(6): 277- 282. URL
	JIA Z, DING Z H, CHEN Y P, et al. Research on event extraction method for judicial data. Computer Engineering and Applications, 2023, 59(6): 277- 282. URL
25	DEVLIN J, CHANG M W, LEE K, et al. BERT: pre-training of deep bidirectional Transformers for language understanding[EB/OL]. [2023-09-15]. http://arxiv.org/abs/1810.04805.
26	刘赫, 皮俊波, 宋鹏程, 等. 基于混合神经网络的电力调度文本事件抽取方法. 中国电力, 2022, 55(9): 105-110, 120. URL
	LIU H, PI J B, SONG P C, et al. An event extraction method for power dispatching text based on hybrid neural network. Electric Power, 2022, 55(9): 105-110, 120. URL

[1]	李鸿鹏, 马博, 杨雅婷, 王磊, 王震, 李晓. 基于槽位语义增强提示学习的篇章级事件抽取方法[J]. 计算机工程, 2023, 49(9): 23-31.
[2]	杨红菊, 靳新宇. 一个实体关系与事件抽取的通用模型[J]. 计算机工程, 2023, 49(2): 143-149.
[3]	苗佳, 段跃兴, 张月琴, 张泽华. 基于CNN-BiGRU模型的事件触发词抽取方法[J]. 计算机工程, 2021, 47(9): 69-74,83.
[4]	柳亦婷, 李培峰. 基于局部实体特征的事件触发词抽取[J]. 计算机工程, 2019, 45(11): 213-217,224.
[5]	陈斌,周勇,刘兵. 基于卷积双向长短期记忆网络的事件触发词抽取[J]. 计算机工程, 2019, 45(1): 153-158.
[6]	梁月仙,陈自岩,王洋,张跃,郭智. 基于时空分析的突发事件检测方法[J]. 计算机工程, 2018, 44(5): 7-13.
[7]	侯立斌, 李培峰, 朱巧明. 基于CRFs和跨事件的事件识别研究[J]. 计算机工程, 2012, 38(24): 191-195.

选择文件类型/文献管理软件名称

选择包含的内容