In the existing folksonomy system, users can be allowed to add any social tags to the resources, but tags are fuzzy and redundancy in semantic, which make it hard to obtain the required information for users. An optimized tag recommender algorithm is proposed to solve the problem in this paper. First, based on the motivation theory, the recommender system uses the model given to calculate the user retrieval motivation before searching information. Second, we use the results in first step to distinguish the user’s type and then cluster the resources tagged according to users who have the similar retrieval motivation with k-means++ algorithm and recommend the most relevant resources to users. The experimental results show that our proposed algorithm with user retrieval motivation can have higher accuracy and stability than traditional retrieval algorithms in folksonomy system.