Improved Latent Factor Model in Movie Recommendation System

In the movie recommendation system, the commonly used recommendation algorithm is the collaborative filtering algorithm. When the sparseness of user behavior data increases, the time and space complexity of the algorithm greatly increase. Latent factor model based on nonnegative matrix factorization is an effective way to solve the sparseness problem. This article focuses on the application of latent factor model in movie recommendation system, and improve latent factor model to overcome the shortcoming that it cannot give the recommendation explanation. In our model, we represent latent factors by characteristic nouns, and then give the recommendation explanations according to the descriptive latent factors. We complete the improved latent factor model on MovieLens datasets.