For higher text classification precision, a general fusion classification model and algorithm are proposed, which based on model theory of information fusion, adopting multi-Media information on the network. The model includes two layers, one is feature layer, which deals with different Media information with different classification algorithm, and inputs the classification results into the higher layer fusion centre separately. The other is fusion layer, which deals with the results from the feature layer, and concludes the final classification result. The experiment expresses the fusion model can improve the text classification precision effectively.