Design And Implementation Of Heat Prediction System For Large-scale Network Event

With the popularity of the Internet,Internet users can express their views on a certain event in social networking sites anytime and anywhere,and a variety of remarks began to flood the Internet.As a result,information grow explosively.However,improper remarks once reached a higher focus on heat,it will have a more adverse impact to the country and individuals,which may cause serious losses.If an event can be predicted earlier before it achieve a higher heat value,the person concerned can guide and controll the event before it reaches a higher heat value.So the real-time prediction of the network heat events has a very important practical significance and practical value.For the above problems,this paper achieved the heat forecast system,which can forecast heat events on four domestic and international social platform,including the well-known social platform Sina microblogging and Baidu Post Bar and foreign social networking platform Twitter and Facebook.The system can predict the real-time heat posted blogs of 5,000 bloggers through 4 channels above.The research contents of this subject include:1.Implement spiders for Sina microblogging,Twitter,Facebook,Baidu Post Bar.Reptile data acquisition side including the direct capture of the page after the analysis and call the official API.2.Porpose a set of reasonable data storage and retrieval programs,it will still be able to maintain high storage efficiency and retrieval speed in the face of data growth.3.Use Spark Streaming flow processing to quickly build predictive models and per-form hot events on events measurement.4.Implement a Java Web site to crawl the crawler directly to the data,extracted from the underlying data to the characteristics and the final results predicted by the form of the chart displayed.This paper monitor more than 5,000 users of the blog through the four channels for a period of 3 months of heat forecast and found that the system has a high stability and faster data capture and modeling calculation speed.The prediction of the heat events of the network is given in a short time.