Collaborative filtering recommendation algorithm for reinforcement learning optimization LFM

Abstract

The invention discloses a collaborative filtering recommendation algorithm of reinforcement learning optimization LFM. The collaborative filtering recommendation algorithm comprises a training part and a prediction part. The training part mainly comprises two steps: firstly, carrying out model training on a training set by adopting an LFM algorithm, and secondly, training a reinforcement learning model; the prediction part mainly comprises two steps: step 1, firstly, obtaining a prediction score value according to an LFM recommendation model; and 2, optimizing the prediction score by using an optimization model. According to the collaborative filtering recommendation algorithm of the reinforcement learning optimization LFM, the influence of a time effect on recommendation performance is considered, modeling is performed on users, scores, books and time through a Markov decision process, the recommendation algorithm is optimized by using a reinforcement learning Q-learning algorithm, the recommendation effect is improved, and prediction is completed.

Description

technical field [0001] The invention relates to the technical field of computer engineering, in particular to a collaborative filtering recommendation algorithm for strengthening learning and optimizing LFM. Background technique [0002] With the rapid development of the Internet, the Internet of Things and cloud computing, the amount of information in the real society is increasing rapidly. Faced with a variety of information, how to obtain personalized services has become an urgent requirement for people. Personalized recommendation analyzes users' behavior preferences through various recommendation algorithms, effectively filters unnecessary information for users, and then provides personalized recommendations for users. At present, personalized recommendation has been widely used in social networking, news, music, movie and book systems, such as Kugou music, Taobao product recommendation, and e-book recommendation. [0003] Collaborative Filtering (Collaboration Filteri...

AI Technical Summary

Problems solved by technology

[0006] Based on the above deficiencies in the prior art, the technical problem solved by the present invention is to provide a collaborative filtering recommendation algorithm for reinforcement learning optimization LFM, using latent semantic model LFM as the core algorithm of collaborative filtering to solve the problem that the time effect will affect the recommendation performance , based on the impact of time effects on recommendation performance, model users, ratings, books, and time through the Markov decision process, and use the reinforcement learning Q-learning algorithm to optimize the recommendation algorithm, improve the recommendation effect, complete the prediction, and effectively solve the problem. The problem of time influence is not considered in book recommendation prediction

Images