Research On Emerging Hotspot Event Detection On Microblog

With the development of social networks, users urgent need a simple and efficient information publication and Obtain way, so the emergence of microblogging has become an inevitable trend. The features real-time performance and convenience of microblogging makes its rapid growth, and its will produce a lot of text information everyday which have practical value, some of it descriptions and comments currently living hot events. Microblogging platform often listed some hotspot topics or events on the first page and recommended it to the users, but it isn’t given more real-time hotspot events of every day. These events often can not satisfy users needs of emerging events,and if you want to find some emerging events which you interestting, you can only read the related posts to learn the topic details,but in this internet information explosion times,it must bring great inconvenience.A research hotspot is focus on accurately and efficiently mining happening hotspot events from microblogs in recent years.In this paper,our research is detection emerging hotspot events in microblogs, detection emerging hotspot events have practical application, some emerging events,such as natural disasters, infectious diseases and attacks if we can detected it at an early time can effectively help government or singel person prevention and treatment the event on time,and decrease unnecessary harm and loss.This paper proposes an approach which can detection early emerging hotspot events based on microblog influence, the influence of microblog calculation by user’s number of fans and microblog forwards, comments times. Firstly is preprocess microblogging datas,and obtained the feature vectors.Then clustering microblogging messages to events by use hierarchical clustering, clustering results is the candidate set of events, during the clustering using a sliding window control datas into the system. Finally, use microblog influence and the publication number of messages calculation emerging hotspots score of each event, and extract keywords constitute event abstract at the same time,after complete all previous step, the top-k ranked emerging hotspot events will recommended to the users.Our approach evaluated by using sina microblog data in real-world, the experiments prove that the proposed approach can effectively detect emerging hotspot events at the early time in microblog, and have some practical application value.