The application of the text clustering technology to the customer service records in Mobile Communications Industry was emphasized. The notional model of knowledge classification based on the text clustering and a concrete executive approach is proposed. The Sets Concept was used to make the definition of the customer knowledge in Mobile Communications Industry. Vector Space Method (VSM) was introduced during knowledge transformation and matrix construction. And TF-MI function is proposed to calculate the weight of the characteristic words. Different approaches to clustering were compared, and hierarchical clustering analysis was explored. The result of the clustering was analyzed and verified by using the four rules presented by Ma Qingguo.