Design And Implementation Of The Analysis System Of Family Viewing Characteristics

With the rapid development of the Internet of Things,the concept and products of smart home also rise rapidly.As one of the most important consumer products in home life,smart TV has also developed rapidly.Smart TV makes it possible to obtain massive video data,which greatly enriches people's daily life.However,although rich TV programs provide people with more choices,they also bring difficulties to TV users' choices.Users are naturally unwilling to spend their precious leisure time on the selection of TV programs,and a personalized recommendation system came into being.Personalized recommendation system is to recommend TV programs that TV users may be interested in on the basis of understanding the basic needs of TV users.To a certain extent,it can help TV users to quickly find TV programs that they are interested in from a large number of TV programs.However the current most personalized recommendation research treat the TV user as an individual when understanding the basic needs of TV users,ignoring the medium of television's biggest characteristics-for families with members greater than or equal to 1.There may be an extreme problem of zero recommendation accuracy for a member of the family in this kind of circumstance.Aiming at this problem,this paper proposes to analyze family viewing characteristics from the two dimensions of time and program,to understand the time characteristics of watching TV and the characteristics of program in different time periods.as a result,the personalized system can make personalized recommendation of TV programs in different time periods and improve the product experience of TV users.The main research contents of this paper are as follows:Firstly,this paper puts forward the basic method of behavior definition and characteristic analysis of TV users' viewing.First of all,this paper constructs a state transfer model with TV programs as the object,and gives the definition of TV users' viewing behavior based on this model.In this paper,three basic behavior states and three basic behavior actions are used to define TV users' viewing behavior.Then based on this definition,this paper takes the viewing behavior data provided by the operators of a province as the source data,and describes in detail the basic methods to analyze the characteristics of family viewing from two dimensions of time and program.Among them,when analyzing the characteristics of family viewing from the program dimension,this paper proposes an optimization scheme suitable for the short-text oriented LDA model in the scene of this topic,and determines the probability distribution of TV program topics through the optimized LDA model.Second,this paper completes the design and implementation of the system.Firstly,the basic requirements of the system,including functional requirements and non-functional requirements,are defined from the application scenarios of the system.Then,based on the basic requirements of the system,this paper designs the overall architecture of the system and the technical architecture of the system,divides the functional modules of the system,and gives the database table structure design of each functional module.Finally,this paper realizes the system functions one by one.Thirdly,this paper completes the test of the system.In the process of system implementation,this paper has conducted a series of test to the system,including the algorithm performance test,system function test and system function test,the algorithm performance test verifies the feasibility of the LDA model optimization,system function test verifies the correctness of the system function,while non-functional test verifies the system can satisfy non-functional requirements.