Research On Personalized Recommendation Based On Spatio-temporal Data And Social Relations

In recent years,the rapid development of IT industry,and information technology in life has also been widely used,especially the emergence of mobile smart devices has brought a lot of convenience to people’s life,the people can get a lot of information.Based on social networks makes people become nearer,interpersonal communication costs are getting lower.At the same time,the application of positioning technology is gradually mature,and many mobile terminals are connected to the network,and then formed a huge system of social network,through social networks,people can understand all kinds of information more comprehensively.Compared with the traditional social networks,location-based social networks have more The advantages,the most obvious is that the geographical factors into it,allowing the user to sign in the network,the virtual world and the physical world will fully integrate the user can very convenient to their geographical location to share.Modern social networks are based on the position on the basis of a large amount of spatial information are friends included in which the user interest mining brings a lot of convenience,at the same time to achieve the position of personalized recommendation,which provides the basic data better.However,it is necessary to point out that due to the large amount of data,so the user to find the information they want are difficult.With this social network There,the user will get a better experience.So far,the position of social network and has many fields together,and electronic commerce and the O2 O is a typical representative,with these industry contacts to create greater economic value.In view of this,in recent years the personalized recommendation technology for position of social networks has become the focus of the study.Overall,at present domestic research on social network recommended position is relatively backward,most research is on the history of user location information,so ignore the interaction between online and offline mode,so the research is not rich enough.In this paper based on space-time number On the basis of mining,the personalized recommendation of location-based social network is deeply discussed.This paper mainly studied the following aspects:(1)data collection and preprocessing.Firstly,collecting real data from a large number of time,space and other aspects related to the statistical analysis of these data,find the user’s behavior,and analyzes the main influencing factors of mobile users(2)according to the subject.The position of the user preference,a method for measurement and calculation of preference.(3)when calculating the user preference considering the factor of time,and the concept of time preference similarity are described,from the angle of time decomposition of the user and the sign matrix,matrix was filled,effective To avoid the occurrence of sparse matrix problems.(4)the time preference and social relations similar well together,and put forward the method of weighted similarity measure,verify the feasibility in the position recommended.Through experimental evaluation and reference to similar effects than after the discovery,the proposed method can a good solution to the cold start problem,can give full play to the role of the position recommended.