【Purpose/significance】In view of the low accuracy of recommendation caused by the less user tags, lack of tag semantics

and the high dimensional sparsity of user-tag matrix in the current microblog content recommendation methods, a personalized recom⁃

mendation method based on social tag mining is proposed.【Method/process】First, aiming at the problem of less user tags, user tags are expanded with the help of microblog content to form the initial user-tag matrix to represent users' interests and preferences;then, for the lack of tag semantics, through the label semantic mapping, mining and semantic relevance, the tag semantic similarity matrix is

constructed to update the user-tag matrix, making the updated user - tag matrix integrate into the semantic correlation between tags. It can solve the problem of users - tag high-dimensional sparse matrix, and better represent users' interests.【Result/conclusion】Experi⁃mental results show that this method is better than the traditional recommendation method based on social tags, and can effectively im⁃prove the accuracy of microblog content recommendation.【Innovation/limitation】In the future, we will focus on the impact of social re⁃lationship between users on personalized recommendation, and explore personalized recommendation method of microblog content inte⁃grating user social relationship, so as to achieve more accurate recommendation.