If the learning resource recommendation method fully considers the efficiency improvement of college students' online independent learning, it can save college students' learning energy and improve their learning quality, but also can promote students to develop such learning attitudes as active participation, willingness to explore and frequent practices. Therefore, this article studies the learning resource recommendation method for improving the efficiency of college students' online independent learning. This article quantitatively evaluates the learning efficiency of college students in the process of online independent learning, and takes the evaluation results as a reference for learning resource recommendation methods. The capsule network combined with self-attention routing algorithm is used to represent the various demands of college students' online independent learning with multiple vectors, and the sequence information layer based on Transformer model is set up to construct the learning resource recommendation model. Experimental results verify the effectiveness of the model.