Personalized Collaborative Filtering Recommendation Algorithm Based On Label Weight

The huge amount of data generated by the rapid development of Internet technology makes it difficult for users to find valuable information quickly and efficiently.In order to help users solve this problem,the recommendation system has been extensively studied.Because of its simplicity,easy implementation and good recommendation effect,the collaborative filtering algorithm has become the mainstream recommendation technology.However,its recommendation performance is often affected by the sparse user operation data.Failure to explain the reasons for the recommendation also affects the user's experience.Therefore,this paper proposes a personalized collaborative filtering recommendation algorithm based on tag weights for the problem of sparse data and poor interpretability in traditional collaborative filtering algorithms.First,using TF-IDF method to calculate user's label weight matrix based onthe user-item scoring matrix and the project label,and then three different methods are used to carry out experiments:1.A personalized collaborative filtering recommendation algorithm combined with a linear regression model.Constructing thelinear regression model based on the userlabel weight matrix and the user-item scoring matrix,and the gradient regression method is used to minimize the value of the linear regression cost function to obtain the project label.Then,the user and project label weight matrix are substituted into the linear regression model to obtain the user's predicted scores for all unrated items.Using the SlopeOne algorithm principle,calculate the difference between the predicted score and the actual score,and the predicted result is adjusted to obtain the final predicted score.Sort the results and recommend Top-N projects to target users.2.Collaborative filtering recommendation algorithm based on nearest neighbor users.Firstly,according to the weight of user label,the nearest neighbor propagation algorithm is used to cluster the users,and the nearest neighbor users of the target users are found in the clustering results.Then,according to the project rating of the neighboring user generated the recommendation for the target user.3.Mix the results of the above two algorithms to generate a score prediction for the target user and generate recommendations.The algorithm transforms user scores and project labels,which can reduce the dimensions of the original matrix and solve the problem of sparse user score data.Experiments using MovieLens dataset show that the algorithm's recommendation accuracy is significantly improved than the traditional collaborative filtering algorithm.And the recommended results are interpretable and can meet the individual needs of users.