To solve the problem of overfitting caused by excessively sparse rating matrices in matrix decomposition, we propose a matrix factorization recommendation algorithm that fuses tags and time information (TTMF). Labels and time information are combined to enrich a single data source and to mitigate overfitting problems in matrix factorization. First, we define the user's tag preference value and item tag relevance defined by rating data and tag information, which represent the user's interest in tags, the relationship between tag information and items, respectively, and we add time information to indicate changes in user interest over time. The user item, user tag, and item tag matrix models are matrix-decomposed by gradient descent to complete recommendations. Experiment results on the MovieLens dataset show that the root mean square error of the TTMF algorithm is reduced by 7% compared with the traditional method latent factor model, and the TTMF algorithm has a better recommendation effect.