Association Analysis Of User Location,Social Behavior And Browsing Behavior Based On Large-scale Network Traffic

The rapid development of mobile Internet has brought great convenience to people's lives.Meanwhile,a lot of data have been generated.Mining more valuable information from these data is of great significance to improving user experience and promoting social development.The study of human mobility has important research value in many fields such as nature and society.Firstly,this thesis analyses the user's movement characteristics,and then puts forward the rules for identifying important locations,such as home and workplace.Based on the user's important locations,this thesis further studies the division of urban functional areas and commuting distance.These studies are helpful to understand the level of urban development and provide reference for urban planning.Trajectory similarity can reflect the relationship between users.In this thesis,a trajectory similarity algorithm is proposed,which is more suitable for the position data fr-om base station.There is a correlation between user mobility and browsing behavior.Firstly,we analyze the characteristics of user's browsing behavior to find out their browsing habits and the popularity of stores.Then,from the point of view of statistical analysis,the distance between users and the shops they browse is quantitatively analyzed.We explore the changes of user's browsing behavior with distance and discover the geographic aggregation of browsing behavior.Finally,the relationship between user location and browsing behavior is explored from the perspective of multilayer network.Multilayer network is helpful to discover phenomena that can not be detected by conventional statistics.Based on the user location and browsing URL,this thesis proposes a series of novel features and train classification model.Results show that the proposed features can reflect the friendships even in general social network and improve prediction performance of traditional features,such as common friends.It provides new ideas for the improvement of social friend recommendation.In addition,our resear-ch is based on the massive real traffic data of a city in northeastern China,which makes the experimental results more reliable and has a strong guiding significance for research and application.