军事事件主题检测与抽取方法  

Military Event Theme Detection and Extraction Method

作　　者：傅浩 刘姗姗 李佳蔚 张志政[3] 张慧 FU Hao;LIU Shanshan;LI Jiawei;ZHANG Zhizheng;ZHANG Hui(Unit 31153 of PLA,Nanjing 210016,China;The 63th Research Institute of National University of Defense Technology,Nanjing 210007,China;School of Computer Science and Engineering,Southeast University,Nanjing 211189,China)

机构地区：[1]解放军31153部队,南京210016 [2]国防科技大学第六十三研究所,南京210007 [3]东南大学计算机科学与工程学院,南京211189

出　　处：《指挥信息系统与技术》2024年第1期76-81,共6页Command Information System and Technology

摘　　要：针对主题汇聚功能,为挖掘军事事件潜在语义关联、抽取事件主题知识,提出了一种面向军事事件的主题检测与抽取方法。首先,采用2种聚类策略从不同层面进行事件聚类,以完成事件主题检测;然后,结合词聚类分析与关键词排名算法,给出各合理事件主题表示,并将其作为事件主题知识进一步丰富军事事件知识图谱。Aiming at the topic aggregation function,and to mine the potential semantic association of military events,and to extract event topic knowledge,a theme detection and extraction method for military events is proposed.Firstly,two clustering strategies are used to cluster events from different levels to complete event theme detection.Then,combined with word clustering analysis and keyword ranking algorithm,the reasonable event theme representation is given,and it is used as the event theme knowledge to enrich the military event knowledge graph further.

关 键 词：知识融合 事件抽取 实体对齐 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]