|本期目录/Table of Contents|

[1]邓云峰,冯永康,王双燕.应急决策文本的多维语义挖掘方法*——基于TF-IDF和PMI的技术框架[J].中国安全生产科学技术,2025,21(5):36-45.[doi:10.11731/j.issn.1673-193x.2025.05.005]
　DENG Yunfeng,FENG Yongkang,WANG Shuangyan.Multidimensional semantic extract method for emergency decision-making texts:a technical framework based on TF-IDF and PMI[J].Journal of Safety Science and Technology,2025,21(5):36-45.[doi:10.11731/j.issn.1673-193x.2025.05.005]
点击复制

应急决策文本的多维语义挖掘方法*——基于TF-IDF和PMI的技术框架()

《中国安全生产科学技术》[ISSN:1673-193X/CN:11-5335/TB]

卷:
21
期数:
2025年5期
页码:
36-45
栏目:
学术论著
出版日期:
2025-05-30

文章信息/Info

Title:
Multidimensional semantic extract method for emergency decision-making texts:a technical framework based on TF-IDF and PMI
文章编号:
1673-193X(2025)-05-0036-10
作者:
邓云峰; 冯永康; 王双燕
（1.中共中央党校（国家行政学院） 应急管理研究院，北京 100089；
2.中国地质大学（北京），北京 100083）
Author(s):
DENG Yunfeng;  FENG Yongkang;  WANG Shuangyan
(1.Party School of the CPC Central Committee (National Academy of Governance),Beijing 100089,China;
2.China University of Geosciences Beijing,Beijing 100083,China)
关键词:
应急决策文本; 词频-逆文档频率; 点互信息; 关联性分析; 复杂网络; 依存句法分析
Keywords:
emergency decision-making texts;  TF-IDF;  PMI;  correlation analysis;  complex network;  LTP
分类号:
X913.4
DOI:
10.11731/j.issn.1673-193x.2025.05.005
文献标志码:
A
摘要:
为了解读领导干部应急决策部署中内含的多维语义信息，了解其相关决策行为特征，进而保障突发事件的应对效果，本文提出基于TF-IDF和PMI的自然语言处理技术框架，挖掘应急决策文本中的多维语义信息，分析相关内容的关联性。首先通过Jieba分词和LTP平台的依存句法分析，挖掘文本中的目标和行动信息，利用TF-IDF算法和词云图展示关键行动，然后通过PMI构建复杂网络，揭示行动间的关联性和决策偏好。研究结果表明：结合应急行动分类体系，TF-IDF算法能精确提取文本中目标和行动信息，以频次反映行动的受关注程度，为理解决策者的决心和意图提供支持；PMI和改良PMI方法能有效挖掘行动的共现关系，揭示行动之间的关联性和决策偏好，其中PMI方法适合分析行动之间的平均相关性，而改良PMI方法能识别出低频高权重的行动关联。研究结果可为分析决策行为特征，细化实化应急决策部署提供可扩展的支持性工具。
Abstract:
In order to interpret the multidimensional semantic information contained in emergency decision-making and deployment of leading cadres,and to understand the characteristics of their decision-making behaviors for ensuring emergency response effectiveness,this study proposes a natural language processing framework integrating TF-IDF and PMI to extract multidimensional semantic information from emergency decision-making texts and analyze their correlations.Firstly,Jieba word segmentation and dependency syntactic analysis via the LTP platform are employed to extract objectives and action information from texts,key actions are visualized using TF-IDF algorithms and word clouds.Subsequently,complex networks are constructed with PMI to reveal correlations between actions and decision-making preferences.The findings demonstrate that combined with emergency action classification systems,the TF-IDF algorithm could effectively extract textual objectives and actions,reflect operational priorities with frequency metrics,thereby supporting the interpretation of decision-makers’ resolve and intent.Both PMI and enhanced PMI methods can successfully identify action co-occurrence relationships and reveal the correlations between actions and decision-making preferences.In this process,PMI reveals average correlations between actions,the enhanced PMI method detects low-frequency yet high-weight action associations.In conclusion,this technical framework provides an extensible analytical tool for characterizing decision-making behaviors and refining emergency deployment strategies.

参考文献/References:

［1］钟开斌.应急决策——理论与案例［M］.北京:社会科学文献出版社，2014:148-237.
［2］李荣志，蔡建淮.领导干部突发事件应急意识的现状调查与分析——基于某省的问卷调查［J］.中国行政管理，2013(1):76-79.LI Rongzhi，CAI Jianhuai.The Current Status and Its Analysis of the Leading Cadre Emergency Consciousness［J］.Chinese Public Administration,2013(1):76-79.
［3］郝阳,邓云峰,程五一.基于“情景－应对”数据的高层官员应急决策行为分析框架［J］.中国安全生产科学技术，2019,15(3):5-10.HAO Yang,DENG Yunfeng,CHENG Wuyi.Analysis framework for emergency decision-making behavior of senior officials based on scenario-response data［J］.Journal of Safety Science and Technology,2019,15(3):5-10.
［4］郝阳,邓云峰,程五一.基于演练决策文本分析的决策者分类及应急决策差异研究［J］.中国安全生产科学技术，2020,16(4):12-18.HAO Yang,DENG Yunfeng,CHENG Wuyi.Research on decision-maker classification and emergency decision difference based on text analysis of exercise decision-making［J］.Journal of Safety Science and Technology,2020,16(4):12-18.
［5］苏婧琼,苏艳琼.基于LDA和TF-IDF的关键词提取算法研究［J］.长江信息通信,2024,37(1):78-80.SU Jingqiong,SU Yanqiong.Research on Keyword Extraction Algorithm Based on LDA and TF-IDF［J］.Changjiang Information & Communications,2024,37(1):78-80.
［6］代钰琴,徐鲁强.基于语义分析的改进TF-IDF算法［J］.西南科技大学学报,2019,34(1):67-73.DAI Yuqin,XU Luqiang.Research on Improved TF-IDF Algorithm Based on Semantic Analysis［J］.Journal of Southwest University of Science and Technology,2019,34(1):67-73.
［7］李峰,陈杰,崔家瑞,等.基于海上救援的稀疏张量局部LDA特征提取算法［J/OL］.计算机应用与软件,1-10［2024-11-25］.http://kns.cnki.net/kcms/detail/31.1260.TP.20240925.1035.004.html.
［8］陈述,孙孟文,陈云,等.基于无监督LDA的水电工程施工安全事故致因分析［J］.中国安全科学学报,2023,33(10):79-85.CHEN Shu,SUN Mengwen,CHEN Yun,et al.Causal analysis of construction safety accidents in hydropower projects based on unsupervised LDA［J］.China Safety Science Journal,2023,33(10):79-85.
［9］李琰,刘珍,陈南希.基于矿工大数据的不安全行为主题挖掘与语义分析［J］.煤矿安全,2023,54(9):254-257.LI Yan,LIU Zhen,CHEN Nanxi.Topic mining and semantic analysis of unsafe behavior based on miner big data［J］.Safety in Coal Mines,2023,54(9):254-257.
［10］谢红.基于词频比的改进Jaccard系数文本相似度计算［J］.内江科技,2021,42(8):27-28.
［11］孙玉岭,孟超,王景春,等.铁路建设管理行为与安全事故致因关联性分析［J］.中国安全科学学报,2018,28(增刊1):71-76.SUN Yuling,MENG Chao,WANG Jingchun,et al.Correlative analysis on causes of railway constructionmanagement behavior and safety accidents［J］.China Safety Science Journal,2018,28(Supplement 1):71-76.
［12］郑秀梅,田晓康,柳青,等.通用航空事故致因文本挖掘和社会网络分析［J］.安全与环境学报,2024,24(2):602-609.ZHENG Xiumei,TIAN Xiaokang,LIU Qing,et al.Text mining and social network analysis of general aviation accident causes［J］.Journal of Safety and Environment,2024,24(2):602-609.
［13］李珏,刘焕.基于HFACS的施工触电事故致因的关联性分析与干预对策［J］.长沙理工大学学报(自然科学版),2019,16(2):36-41,48.LI Jue,LIU Huan.Association of causes and intervention countermeasures of electricshock accidents based on HFACS［J］.Journal of Changsha University of Science and Technology (Natural Science),2019,16(2):36-41,48.
［14］刘永强,谢成斌,高阳,等.基于文本挖掘的水利工程施工人员不安全行为影响因素研究［J］.中国农村水利水电,2024(6):150-155.LIU Yongqiang,XIE Chengbin,GAO Yang,et al.Research on the factors influencing unsafe behaviors of waterconservancy engineering construction personnel based on text mining［J］.China Rural Water and Hydropower,2024(6):150-155.
［15］骆逸欣,严志科,尹红炼.改进的点互信息微博情感分类方法［J］.电脑与信息技术,2017,25(2):14-15,19.LUO Yixin,YAN Zhike,YIN Honglian.The Weibo emotion classification method based on an extended semantic orientation pointwise mutual information［J］.Computer & Information Technology,2017,25(2):14-15,19.
［16］范庆春,匡皓松,谢飞.融合词向量和点互信息的领域情感词典构建［J］.阜阳师范大学学报(自然科学版),2021,38(3):73-80.FAN Qingchun,KUANG Haosong,XIE Fei.Domain sentiment lexicon construction combining word embedding and pointwise mutual information［J］.Journal of Fuyang Normal University(Natural Science),2021,38(3):73-80.
［17］CHE W,FENG Y,QIN L,et al.N-LTP:An open-source neural language technology platform for Chinese［C］//Empirical Methods in Natural Language Processing.Online and Punta Cana,Dominican Republic:Association for Computational Linguistics,2021:42-49.

相似文献/References:

备注/Memo

备注/Memo:
收稿日期： 2025-03-31
* 基金项目： 国家重点研发计划项目（2022YFC3005701）
作者简介： 邓云峰，博士，研究员，主要研究方向为公共安全和应急管理，应急演练与应急决策。
通信作者： 王双燕，博士，讲师，主要研究方向为指挥决策、信息共享等。
更新日期/Last Update: 2025-05-26