摘要
为了研究微博的事件感知与脉络呈现方法,以Twitter为研究对象,对现实生活中发生的事件进行提取并呈现事件发展的过程.对微博的处理分为事件感知阶段和事件脉络呈现阶段.在事件感知阶段对原始微博进行过滤分析,去除冗余信息,并得到与事件相关的微博集.在事件脉络呈现阶段采用基于图结构的方法,将微博之间的关系转换成图中结点之间的关系,寻找图中的关键结点作为关键微博,并连接关键结点,最终得到在时间和内容上连贯的事件脉络.实验结果表明:所提出的方法能呈现事件的发展过程,也能体现事件发展的多样化.
The event sensing and vein presenting problem with the data from Twitter was investigated to extract real-life events and the development of the event and finally present a comprehensive event vein.Microblogging process was made up of two main modules,including event sensing and event presentation.The event sensing module processed raw microblogs,filtered redundant information and extracted the ones associated with the event.The event presentation module presented the event vein based on the relationship between microblogs.Next,an effective approach based on the graph structure was proposed to transform the relationship between microblogs to the relationship between nodes,each of which in the graph represented a microblog.Key nodes was identified in the graph,and then linked with edges.Finally,the event vein that ensured both temporal and content coherence was generated.Results of experiments over a real dataset collected from Twitter show that our approach to generate the event vein is effective and also can reflect the diversity of events.
出处
《浙江大学学报（工学版）》
EI
CAS
CSCD
北大核心
2016年第6期1176-1182,共7页
Journal of Zhejiang University：Engineering Science
基金
国家"973"重点基础研究发展规划资助项目(2015CB352400)
国家自然科学基金资助项目(61332005
61373119
61222209)
关键词
微博
事件感知
事件脉络
图挖掘
microblogging
event detection
storyline
graph mining
