Research on Intelligent Personalized Recommendation of Library Based on Matrix Decomposition Implicit Semantic Model

Abstract

In order to quickly select a book from a large number of books, a comprehensive method is proposed with an implicit semantic model based on matrix analysis and time effect according to the cognitive characteristics of university readers in different learning periods. This method uses the random gradient descent method to calculate the customer-project evaluation matrix. In the method, a new treatment method is provided for cold start. The absolute error (MAE) and the mean square relative error (RMSE) of the evaluation index are used to test the correctness of the information provided by the proposed method. The feasibility and effectiveness of this method are confirmed by the actual data.