The current personalized learning recommendation has been proposed to improve the efficiency of online teaching, but the current personalized learning recommendation algorithm still has problems such as large redundancy of learning resources, small improvement of learning performance, learner differentiation not fully satisfied and limited recommendation accuracy. To address the above problems, this paper proposes a personalized learning recommendation algorithm based on test questions. The algorithm combines linear regression and EM algorithm to calculate each student’s mastery of each knowledge point. The test practice is used as the basis for pushing other personalized learning resources for students in combination with the knowledge map. The experimental results show that the algorithm has higher recommendation accuracy, recall rate compared with existing algorithms.