Research On Recommendation Algorithms Of Personalized Intelligent Recommendation System For TV Home Users

The Television,which was invented in 1925,has greatly changed people's daily life,include people's leisure and entertainment.As a result,it has been an absolutely necessary part of people's life.As we entered the 21 th Century,radio and TV operators can interact with many family users to achieve real-time information.It makes the full and personalized product marketing and paid services become a reality.Intelligent recommendation,as a new machine learning algorithm,has been widely applied in the internet industry,however,to find it in the field of radio and television is unusual.In order to propose an effective algorithm for radio and television recommendation,firstly,we have learned the achievements of previous research on the Algorithm of Intelligent Recommendation,and the application in the filed of radio and television.Secondly,we compared the recommendation algorithm of currently mainstream in other product industries,which was content-based on recommendation algorithm,collaborative filtering recommendation algorithm,and association rule-based recommendation algorithm.Finally,we thoroughly explored the principles,advantages and disadvantages of each algorithm,then discussed the feasibility to apply it in the filed of radio and TV program recommendation.Based on a model data of users' information record,which was recorded by a radio and television network operation company,we have made an necessary clean for the data.The Apriori association rule algorithm and the collaborative filtering algorithm are selected to establish the intelligent recommendation model of TV program.This model exploits algorithm to find out the implicit relationship between program products and users,then to recommend products similar to their preference types for the users.In the end,the model effect is evaluated according to the evaluation system used by the machine learning.For the sake of avoiding the limitation of the single model,and to improve the effective of recommendation further,we have made a combination between the Apriori association rule algorithm and the collaborative filtering algorithm,to obtain a hybrid recommendation model in this paper.According to our experience,compared to the single model,the collaborative filtering algorithm can make an effective improvement for the system recommendation.What's more,the collaborative filtering algorithm also can increase the extensibility of the recommendation,to seize the hobby of users,and to achieve the personalized recommendation.Besides,the collaborative filtering algorithm has played a very positive and effective role in enhancing user experience and promoting paid service marketing.