Research And Implementation Of Personalized Movie Recommendation Algorithm Based On Tag

Since the 90s of last century, with the rapid development of the Internet, the movie has also been rapid development, with the emergence of tag information,the tag information can better reflect the contents of a movie. Based on the tag information, this paper proposes the following improvement measures for the shortcomings of the personalized recommendation algorithms:First,we propose a method to extract the weight of movie tags based on word2vec. This method can generate the weight model of the movie according to the user tag information, and propose a movie recommendation method, finally the experiments prove the feasibility of extracting the weight of movie tags.Second, traditional Slope One algorithm does not consider the similarity between movies in the process of scoring, so it will cause the scoring error, this paper introduces the similarity of the weight of movie tags and score to get the target movie's neighbors, and then predict the score of the target movie. Meanwhile, the Slope One algorithm does not consider the similarity between users, in this paper, we propose a method to extract user similarity and add to the Slope One algorithm. Finally, a hybrid Slope One algorithm is proposed by combining with both of two algorithms. The experiments show that the improved Slope One algorithms are more accurate in predicting score.Thirdly, The traditional collaborative filtering recommendation algorithm based on a single score to calculate the similarity will lead to poor recommendation, in this paper, we propose a collaborative filtering recommendation algorithm combined with the similarity of tags weight,at the same time, the traditional collaborative filtering recommendation algorithm data sparse problem will also lead to poor recommendation results, multidimensional sparse matrix is generally used to reduce the sparsity of the matrix through dimensionality reduction. In this paper,singular value decomposition (SVD) is introduced to reduce the dimension of user scoring matrix, and finally,a collaborative filtering recommendation algorithm combined with the weight of movie tags and SVD is proposed. The experiments show that the improved collaborative filtering recommendation algorithm is more accurate in predicting score than the traditional collaborative filtering recommendation algorithms.The experiments prove that combine tag information with personalized recommendation algorithms effectively improves the accuracy of the recommendation, and also-prove the feasibility of personalized recommendation algorithms based on tag.