The invention relates to the technical field of computers, particularly provides a culture resource recommendation system, and aims to solve the technical problem of incapability of realizing effective recommendation of culture resources under the condition of high-dimensional sparse user behaviors. For the purpose, a behavior data collection module of the culture resource recommendation system can collect different types of user behavior data in a preset culture resource platform; a behavior data processing module can obtain user behavior features; a behavior weight assessment module can obtain user behavior weight values; a culture resource clustering module can obtain culture resource types in the preset culture resource platform; a behavior feature calculation module can perform deep learning on user behaviors according to the data; and the culture resource recommendation module can determine recommended culture resources according to a deep learning result by adopting a nearest neighbor collaborative filtering algorithm. Through the system, the effective recommendation of the culture resources can be realized under the condition of the high-dimensional sparse user behaviors.