【关 键 词 】：
Providing personalized learning resources for learners from educational big data is an effective way to reduce learning disorientation and improve learners' online learning experience. The key of realizing accurate learning resource recommendation is to excavate and analyze complicated relationships among the data, including learners, resources, situations, etc. Therefore, tensor theory is introduced to construct a learner-resource fusion tensor. Afterwards, according to the principle of multi-dimensional correlation analysis, the high order singular value decomposition approach is adopted to mine the associations of learners and resources to realize the accurate matching between them. Finally, a contrast experiment is conducted based on 9 groups of data with different scales. The experimental results show that the recommendation performance of multi-dimensional correlation analysis method outperforms that of the collaborative filtering approach maturely applied to the recommendation field, and the larger the data scale becomes, the better the recommendation performance. It is implied that the multi-dimensional correlation analysis method is beneficial to the accurate recommendation of personalized learning resources in age of big data, to improving the quality of online education and enhancing personalized learning effect.