Visual Analysis Of Microblog Data

Nowadays,Social media develop very rapidly base on the Internet.Take Twitter for example,people who loves to use Twitter,pushes all sorts of novel idea,and shares links of interesting news,pictures,music,videos joyfully every day,these tweets retweeted again and again,spread around the world rapidly.However,it is difficult to effective retrieval of relevant messages from microblog data,because it’s mass,dynamic,complex and various.The traditional data analysis approach will not work.This paper present an interactive visual analysis system,using twitter data as the research object,analyze the data by machine learning algorithm,combining with visualization tools,analyze the data visually and support the human-computer interaction,help the user to filter the mass and the changing data,monitoring their topic of interest.On the aspects of data analysis,current tools for monitoring microblogs typically filter messages based on user-defined keyword queries.In this paper,classification is realized by using support vector machine,user can select classifier which has been created to filter tweets through a visual interface.On the aspects of visualization,the system is realization based on Web technology.User can observed the dynamic loading of the twitter information intuitively.The way data is presented by scatter diagram or heatmap,support a variety of interactive behavior,such as drag-and-drop,records,tagging,delete,etc.System also use third-party graph library to show statistics alternatively.Direct visualization of the results is difficult to observe for time-varying statistics changes subtly.In this paper,video magnification technology based on pixel is applied to help user to understand the data and analyze the data better.Experimental results show the system can monitor a flow of Twitter stream dynamically,and classify them effectively.With the help of visual technology user can understand the development of events accurately.