Weibo Event Detection And Analysis Based On The Geographic Coordinates

With the further development and application of computer network and web2.0 technology,weibo has changed the traditional form of social network,and quickly became popular in society.Users can not only timely access to the hot spot of social information on the microblogging platform,but also can promote the spread andinfluence of relevant information through the microblogging content sharing and forward. Weibo has a unique form of social network,it is convenient for people in information acquisition, transmission and sharing.The all-round development of mobile intelligent device and wireless network technology,people can issue the social events that took place in the real world to the microblogging platform at any timeany place.Social events are often reflected firstly in weibo content,thus cause theattention of a large number of users and have a great influence.Therefore,weibo event detection has far-reaching significance,it is also one of focuses in the current weibo research.The research target of this paper is to detect significative event information from the multifarious and massive short weibo data by using data mining algorithms,so as to help the user to obtain more valuable events information within a limited time,and can assist the relevant government departments in microblogging public opinion control and social administrative decisions.Firstly,in view of the domestic representative micro-blog website,We through the use of micro-blog open network interface to collect micro-blog data with geographic coordinates.Secondly,we use these datas as the training sample,we use K-means, KNN and decision tree algorithm to construct the geographical regularity of micro-blog characteristics based on the number of the micro-blog published, micro-blog forwarding number,micro-blog comments,user activity and movement strength.Finally,We compare the geographical regularity features with regional micro-blog data characteristics daily,thus it is concluded that whether there are the occurrence of emergency in the region.Based on the use of K-means,KNN and the decision tree data mining algorithm,we put forward the weibo event detection framework based on the geographic coordinates on the basis of data analysis,and perform visual analysis for the development trendof weibo event by using R language.The result of event information by using weibo event detection framework has great reference value.Especially,the feelings of the weibo users and the development trend for the events are worth thinking about and analyzing,and provide reference for a later similar event. Therefore,this paper introduces the theory and technology of information retrieval to the organization and management of weibo eventinformation.In order to facilitate users to retrieve these weibo event information which has detected.In this paper,on the base of Lucene open source search engine,we combine with the mmseg chinese word segmentation technology and jsp dynamic web technology to implement weibo event retrieval system,it can help users understand the content and development trend of the past weibo event.At last,we use experiment to analysis the feasibility of weibo event detection framework.At the same time, we combine with the specific case of the "dust storms" events and realize accurate retrieval of weibo event.