Matrix Factorization Recommendation Algorithm Incorporating Tag Factor

Matrix factorization is a hot spot in the research of recommendation algorithms. Traditional matrix factorization algorithm only learns the user factor and the item factor from rating data, not fully considering the influence of the tag data. Therefore, a matrix factorization recommendation algorithm incorporating tag factor is proposed. This algorithm obtains user's tag preference matrix by considering rating data and tag data comprehensively, and then incorporates item preference tag factor and user's tag preference factor into the matrix factorization recommendation algorithm. In this algorithm, tag-rating sparse coefficient is proposed to better balance the use of latent factors and tag factors in the recommendation process. At the same time, the TF-IDF algorithm is used to calculate the tag factor weight of the item, which reflects the influence of different times of the tag marked on different items. Experimental results demonstrate that the proposed recommendation algorithm can produce better accuracy and performance compared with the traditional matrix factorization algorithm.