摘要
从微博中准确高效地挖掘出正在发生的热点事件是近年来研究的热点。通过综合考虑微博用户的粉丝数量和微博本身的转发、评论次数计算每条微博的影响力,从而提出一种基于影响力的微博新兴热点事件检测方法 IEED(Influence-Based Emerging Hotspot Event Detection)。该方法运用层次聚类将微博帖子聚类为事件集,并提取出事件中的关键词构成事件摘要。通过运用现实生活中的新浪微博数据作为实验数据集来测试所提出的方法,实验结果证明,基于影响力的微博新兴热点事件检测方法(IEED)能在早期高效地检测出微博中的新兴热点事件,具备一定的应用价值。
To accurately and efficiently mine the hot events on occurrence from microblogs is the focus of research in recent years. In this paper we propose an influence-based emerging hot events detection（ IEED） approach by comprehensively considering the fans number of microblogging users and the influence of each microblog calculated from the number of its forwarding and comments. The approach uses hierarchical clustering to cluster the microblogging messages into event set,and extracts the keywords in the events to form event abstracts.We tested the approach presented in the paper by using the experimental dataset set up from Sina microblogging data in real life,the experimental result proved that the influence-based IEED could efficiently detect the emerging hot events in microblogs at early time,and had certain applied value.
出处
《计算机应用与软件》
CSCD
2016年第5期98-101,165,共5页
Computer Applications and Software
关键词
新兴事件检测
微博影响力
聚类
Emerging events detection
Microblog influence
Clustering
