Abstract: To solve the drawback that the precision of the document-level sentiment classification is lower than that of the normal text classification, this paper proposes a semantic weight-based Native Bayesian algorithm for text sentiment classification. First, the words in an emotion dictionary were scored and weighted using a feature selection method. Second, based on the correlation between the distribution of dictionary polar and the document-level sentiment classification, the semantic weight feature was merged into naive Bayesian classification and a new algorithm was achieved. Finally, lots of experiments on some standard Chinese data sets were performed.Resultsshow that this algorithm is better than some existing algorithms on precision, recall, and F₁-measure.