Abstract: A personalized micro-blog search model is proposed by taking into account the user and publisher's modelling. In this search model, the user interest model of the micro-blog theme dimension is constructed by using the theme model and language model, while the user interest model of the user-publisher relationship dimensions is constructed by integrating the features of user-publisher relationship. These user interest models are further integrated to design a single user's micro-blog as a text of the training methods for solving the problems of short texts and sparse notes of the micro-blog. Experiments based on real users' search feedback show that the proposed micro-blog search model can improve the personalized effect of micro-blog search.