Textual documents are created in various forms. Existing work is devoted to topic modeling and the evolution of individual topics, but sequential relations of topics in successive documents created by specific users are ignored. In this paper we introduce Sequential Topic Patterns (STPs) and formulate the problem of mining User-aware Rare Sequential Topic Patterns (URSTPs) in document streams that characterize and detect personalized and abnormal behaviors of users. Such an innovative problem of mining will be solved by using three phases: extraction of probabilistic topics using preprocessing and identify sessions for different users, generating all the STP candidates with (expected) support values for each user by patterngrowth, and selecting URSTPs by making user-aware rarity analysis on derived STPs. Finally our result reflects users’ characteristics.