The Research And Design Of Message Push Service Based On Context And User Behavior

In recent years,the continuous advancement of technology and the rapid spread of mobile smart devices have greatly satisfied people's requirements in production and life.But at the same time,people are facing the increasingly serious problem of information overload.Personalized recommendation technology is considered as one of the effective ways to alleviate this problem.By analyzing user historical data and constructing a user preference model,the personalized recommendation system could actively recommend the most interesting content to the users.At present,mobile network services are on the rise.While we trend to record the details of our lives with mobile devices,a large number of scene data with time and space characteristics have been produced.These data have a wide range of sources,large data scales,and rich information.However,the traditional recommendation technology only focus on mining user behavior information and usually ignores the context information around users,so it is impossible to provide effective recommendation services in the mobile environment.In view of the above problems,this paper studied the impact of context on the recommendation system.Focusing on solving the deficiency of the traditional two-dimensional recommendation algorithm,this system attempts to utilize the context and user behavior information.This system contains two improved personalized recommendation algorithms,and designs and implements a message push prototype system.The paper mainly includes the following four parts:(1)By introducing two methods for processing context information,this paper put forward a method based on the technology of contextual pre-filtering and the improved Pearson Correlation Coefficient.(2)Using the serialization feature of the scenario as the entry point,this paper explores the user life mode.Besides,a method for calculating the similarity of "user-context" is proposed.(3)Using the method of contextual modeling,the relationship between "user-item-context" is deeply explored.In addition,it also adopts the Brightkite dataset as the multiple sets for comparison experiments which confirmed the effectiveness of the algorithm proposed in this paper.(4)A prototype system with functions of data acquisition,analysis,visualization and message pushing is designed,mainly including SDK and data analysis center.