The filing and classification of E-government document in E-government information system are accomplished manually, which restricts the government information technology seriously. To solve this problem, a classification algorithm for E-government document based on support vector machine is proposed. For the general workflow of E-government document circulation in the current information system, a scheme of the auto-classification and archiving based on service is presented so that it is realized in a heterogeneous information system. In this scheme, the original system is reused to create new value as much as possible and change of workflow is responded more rapidly. It is a new solution to the automatic E-government system.