Design And Implementation Of Video Recommendation System Based On Collaborative Filtering

In today’s society,people’s living standards are constantly improving,and in the era of rapid development of information,many behaviors and habits in people’s life are digitalized.In this era of information overload,when facing a large amount of information,everyone must want to spend a little time to find the information they need.As an information query tool in the network era,recommendation system has been developed rapidly since it was proposed as an independent concept in the 1990 s.Personalized recommendation algorithm technology has been widely used in many fields,such as music media websites,online text reading,social networking sites,etc.at present,in most movie websites,they do not grasp the interests and hobbies of moviegoers and then make targeted personalized recommendations,which is limited to recommending several similar movies according to the type of movies they currently watch.After many unsuccessful movie selections,users often choose to close the website and choose other ways to get their favorite movies.In view of the problems existing in the above existing video websites,this paper proposes a film and television recommendation system based on collaborative filtering.The Django framework used can intelligently ensure the security of project data and website.In addition,a recommendation algorithm combining two collaborative filtering algorithms is used,which can improve the accuracy of film recommendation and the loyalty of viewers to the website.The main work and research contents of this system are as follows:(1)Select movielens data set as the basis of this recommendation system,and use Python to compare ratings.dat,movies.dat,and users in the data set DAT files are preprocessed,classified,clustered,reorganized and filtered,and transformed into understandable formats and mining formats.(2)The binary kmeans algorithm is used to obtain similar users,which is a classification clustering algorithm optimized based on the K-means clustering algorithm.By iteratively calling kmeans(k=2),the cluster is divided many times until K users are finally obtained,which not only avoids the algorithm from falling into a local optimal state,but also improves the speed of the algorithm.(3)The improved collaborative filtering algorithm based on usercf and itemcf is combined to optimize the similarity calculation method,and the popular items in the intersection of two user interest lists are punished by using penalty items,so as to obtain K similar users with more accurate similarity.(4)Design the test experiment based on this system,prepare the environment of the experiment,including how to obtain the test data and training data,get the information matrix of the user’s preference record and the item information matrix through the training data,select the similar user with k nearest neighbors to calculate the user’s interest score,and finally compare and analyze the accuracy of usercf algorithm and itemcf algorithm under different movie recommendations.Based on the above technical research,this paper designs and implements a video recommendation system based on collaborative filtering through demand analysis.After the evaluation and system test of the recommendation algorithm after incorporating the penalty factor into the system,the test results and functional requirements meet the expected results.