People use Internet for different purposes e.g. social networking, blogging etc. with respect to their context. This leads to dynamic change in creation and distribution of document streams over the Internet. This would challenge the topic modelling and evolution of individual topics. In this paper, we have proposed Sequential Topic Patterns (STPs) mining over the published user-aware document streams and formulate the problem of mining User- Aware Rare Sequential Topic Patterns(URSTPs) in document streams on the Internet in order to find rare users. They are generally rare and infrequent over the Internet. For URSTPs mining we need to perform three phases: pre-processing to extract topics, generating STPs, determining URSTPs by rarity analysis of STPs. The experiment can be performed on both real times and synthetic data-sets. In the proposed work, we have focused on synthetic datasets.