引用本文: | 张辉,李国辉,徐新文,等.词网络的新闻事件关联建模[J].国防科技大学学报,2014,36(4):169-176.[点击复制] |
ZHANG Hui,LI Guohui,XU Xinwen,et al.Modeling association of news events on term network[J].Journal of National University of Defense Technology,2014,36(4):169-176[点击复制] |
|
|
|
本文已被:浏览 9650次 下载 6837次 |
词网络的新闻事件关联建模 |
张辉1,2, 李国辉1, 徐新文3, 贾立1, 孙博良1 |
(1. 国防科技大学 信息系统与管理学院, 湖南 长沙 410073;2.3. 61226部队,北京 100079;3.2. 国防科技大学 指挥军官基础教育学院, 湖南 长沙 410073)
|
摘要: |
互联网上每天都会报道许多新闻事件,为了挖掘各事件间的关系,提出一种新闻事件关联建模方法。该方法首先利用TF-IEF和相邻词合并策略对事件的相关报道集提取关键词,然后综合多种词共现度量窗口对事件关键词的关联关系建模,构建事件关键词关联网络,最后依据事件间共有关键词的程度建立事件关联模型,从而建立事件关联网络。实验表明该方法能够较准确地提取报道集的关键词,较好地发现事件间的关联关系。 |
关键词: 词网络 新闻事件关联 词共现 |
DOI:10.11887/j.cn.201404029 |
投稿日期:2013-09-05 |
基金项目:国家自然科学基金项目(61170158); 国家部委资助项目; 湖南省自然科学基金项目(12JJ5028) |
|
Modeling association of news events on term network |
ZHANG Hui1,2, LI Guohui1, XU Xinwen3, JIA Li1, SUN Boliang1 |
(1. College of Information System and Management, National University of Defense Technology, Changsha 410073, China;2.3. Unit 61226, Beijing 100079;3.2. College of Basic Education for Officers, National University of Defense Technology, Changsha 410073, China)
|
Abstract: |
There are many news events reported daily on the Internet. An innovative method is proposed to mine event-relations between news. Following an adjacent term-combining strategy, this method primarily utilized a so-called term frequency & inverse event frequency (TF-IEF) model to extract key phrases from the corresponding reports set as to a particular event. Then term co-occurrence windows were employed to calculate the associating degree of every single term pair. This degree is indicative in building event key phrase-networks. Further, two matters were correlated to shape the event relation-network model: (I) common key phrases as mediators within event key phrase-network, and (II) the degree of commonness of key phrases within different observed events. An experiment was conducted to examine the performance of proposed method. The results show that the method can accurately extract key phrases and comprehensively mine associations between events. |
Keywords: term network news event association term co-occurrence |
|
|
|
|
|