Using Weight-Retouching and Under-Sampling SVM Approaches for Text Categorization on Imbalanced Data

More and more textual documents are available, which makes it more difficult to manage text data and to retrieve useful information from document contents. Text categorization is an important way to help resolve this problem, which is an increasingly important field and has been extensively studied. In this paper, we pay attention to the performance of the minority text class and attempt to improve its precision by using some techniques of imbalanced data processing. A weight-retouching and under-sampling Support Vector Machine(SVM) approaches have been taken into account. And it shows that the processing approaches of imbalanced text data by using weight-retouching and under-sampling SVM will make improvement on precision of minority class, while it won´t blemish the global performance.