面向新浪微博签到数据的时空热点事件检测方法

Sina Weibo for the Data of Time and Space Hot Event Detection Method

传统面向文本数据的事件检测方法在处理以微博为代表的社交媒体数据时面临着效率和准确性的挑战。同时,社交媒体数据中富含的位置信息常常不能被有效地识别和利用,这无疑会影响到事件检测的效果。本文基于对已有研究的总结归纳,定义了一类面向微博签到数据的时空热点事件,并提出了一种新的微博时空热点事件检测方法对其进行识别。通过两组实际数据的实验,证明该方法能够有效地从海量的微博数据中挖掘出具有时空特征的热点事件。

A spatio - temporal hot event can be defined as a certain thing happens in somewhere and in a time period. Traditional e- vent detection methods for text data are facing efficiency and accuracy challenges in dealing with social media data, such as Weibo. Meanwhile, they cant use the location information which is rich in social media data effectively. This paper summarized two kinds of location usage modes for current spatial - temporal hot event detection methods on Weibo data. Then, proposed a formalized definition of spatial - temporal event based on Weibo Check in data, and designed new method to identify them. Two experiment results on the real Weibo data show that the new method can detect spatial - temporal hot event from large Weibo data efficiently.