基于新闻文本的事件可视方法研究  

Research on Event Visual Method Based on News Text

作　　者：刘海砚[1] 李佳 刘建湘 陈晓慧[1] 程维应 LIU Haiyan;LI Jia;LIU Jianxiang;CHEN Xiaohui;CHENG Weiying(Information Engineering University, Zhengzhou 450001, China;Unit 32137, Zhangjiakou 054000, China)

机构地区：[1]信息工程大学,河南郑州450001 [2]32137部队,河北张家口054000

出　　处：《信息工程大学学报》2020年第5期601-606,共6页Journal of Information Engineering University

基　　金：国家自然科学基金资助项目(41801313,41901397)。

摘　　要：随着网络媒体的兴起,每天都产生海量的新闻文本数据,这些对事件报道的新闻文本中包含了丰富的信息,如何从这些海量文本中快速获取事件的多维要素成为当前的研究热点。提出了面向新闻文本数据的事件可视分析方法,根据事件模型,利用自然语言处理技术对新闻文本数据进行分词和主题建模,提取事件时空和语义信息;采用日历图可视化时序关系、流地图可视化空间关系,采用主题河流图和节点链接图可视化事件主题演变和关联关系;最后形成多维事件可视分析系统。以中美贸易战时期的相关新闻文本进行实验分析,结果表明,该方法能够快速分析出中美贸易战演变过程,帮助人们快速获取海量新闻文本中感兴趣的信息。With the rise of network media,a huge amount of news text data are produced every day,and these news texts that report on events contain rich information.How to quickly obtain multidimensional features of events from these massive texts has become a current research hotspot.Therefore,this paper proposes an event visual analysis method for news text data.According to the event model,we extract word segmentation and model topic from news text data to extract event spatial and semantic information by natural language processing technology.We visualize the spatiotemporal feature by calendar map and flow map.Theme river map and node-link map visual changing of theme and association of events are used.Finally,the event visual analysis system is formed.In this paper,we use relevant news texts during the Sino-US trade war for experimental analysis.The results show that this visual analysis method can quickly analyze the evolution process of events during the Sino-US trade war and help people obtain information of interest in massive news texts.

关 键 词：新闻文本 时空要素 语义要素 可视分析 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]