Book recommendation method based on personalized recall algorithm LFM

Abstract

The invention discloses a book recommendation method based on a personalized recall algorithm LFM. The book recommendation method based on the personalized recall algorithm LFM is established, the LFM processes a scoring matrix through singular value matrix decomposition SVD to obtain potential features of a user, and missing items are scored according to the potential features; a reinforcement learning method is introduced to improve the accuracy and diversity of book recommendation, a traditional LFM algorithm and a gradient descent method in deep learning are combined and applied to a finally-designed book recommendation model, negative score data of books are introduced on the premise of ensuring an accurate recommendation result, so that the extreme recommendation result is prevented, and the overall diversity of book recommendation is increased. According to the method, a deep reinforcement learning method is utilized, a basic formula is obtained through a gradient descent method, and a loss function and a regularization item are added on improvement of the formula, so that relatively reasonable user vectors and item vectors are obtained without manual intervention, and book results recommended to each user are more accurate.

Description

technical field [0001] The invention belongs to the technical field of recommendation algorithms. The recommendation algorithms are widely used on Internet platforms, and accurate and efficient recommendation algorithms are of great significance to the Internet service industry. Aiming at the field of book recommendation, the present invention has researched a set of matrix decomposition algorithms. Under the conditions of existing data sets and test sets, the personalized recall algorithm LFM is used to accurately recommend books that users like from user behavior data and book data. . Background technique [0002] In the book recommendation system, the recommendation algorithm plays an important role in recommending users' preferences and evaluating system performance. In recent years, with the vigorous development of the Internet service industry, such as e-commerce, small videos, news apps, and film and television apps, all of which involve the use of recommendation alg...

AI Technical Summary

Problems solved by technology

[0006] In view of the shortcomings of the existing collaborative filtering methods item CF and user CF, consider the problem of incomplete conditions affecting the recommendation results

Images