Analysis And Forecast For User Behavior Based On Location Information

With the development of the Internet,Social Network Service(SNS)has become such a Web 2.0 business with the most extensive coverage of users,the greatest impact and the highest commercial value.Mature position technology on mobile terminal greatly improved the quality of Location Based Service(LBS).The birth of a new SNS comes-Location Based Social Network(LBSN).The feature of LBSN is supporting users to record location history.With these information,the virtual world establishes closer ties with the real world.Currently,LBSN has become a new direction in social research.It provides a number of location information with time,space and social relationships for the relevant research.Applications related to LBSN also provides a demand of these research.In this thesis,LBSN is studied with users' check-in history,register locations and friendship relationship.Study establishes a model for user behavior in spatial,temporal and social,according to user check-in action.The thesis presents a personalized Location Recommendation System and gives detail design of that for reality scenario.Based on user characteristics model with multi factors,this system predicts users'potential interest locations,and recommends them to users.This thesis also gives a definition of travel together activity.Based on the online travel together activity data,this thesis proposes a Location Group Recommendation Algorithm facing travel together groups.This algorithm integrates the personalization Location Recommendation System proposed above.The algorithm is achieved on a distributed computation platform,Spark.It is tested with different aggregation methods and the result shows a great improvement on recommendation performance for Location Group Recommendation.Based on theory work,this thesis proposes an application design,providing service on asking remote questions.At last,there are some idea about improvement on theory and experiment.