Recently, more and more researchers realize that it is hard to satisfy the users’ needs of the datadriven unsupervised learning methods in text data analysis. Fortunately, the appearance of domain knowledge provides a bright future for this problem. Domain knowledge, at present, is usually represented as ontology hierarchy (built artificially) or encyclopedia (e.g., online Wikipedia). In this paper, we describe how the domain knowledge is used in processing text mining, review the related work,and point out the challenges in this research direction.