Document Reading - oriented migration method characterized in text categorization

     This paper presents a migration migration based learning solutions, centralized model constructed from the correct training, reduce differences in the training set and test set model in existence, and ultimately improve the classification accuracy of the text. Finally, Chinese spam filtering, Chinese and English web pages classification categories to verify the proposed method. Transfer learning text classification problem there is still much research space, such as the amount of information contained in the part of the words will change over time. In this text categorization, it features not only the characteristic words of a change in distribution, composition and characteristics of the dictionary feature space need to constantly update.