Service system classification method and system

The invention relates to a business system classification method and system. The method comprises the following steps: performing category label labeling on core words through a conditional random field, and capturing an association relationship between the core words to provide more semantic information for the core words; a theme matrix is extracted through non-negative matrix factorization, semantic representation of the core words is further enhanced, and theme association information about the core words is provided; and combining the two features into feature vectors of the business system to more comprehensively describe semantic features of core words of the business system, so that deep data analysis is facilitated, and multi-level information organization is realized. According to the combined feature vectors, clustering classification is carried out based on calculation of an initial clustering center and a service system similarity distance and re-clustering so as to carry out similar division on