A New Method of Text Feature Selection for Knowledge Discovery

Abstract

a hearingme@yeah.net Keywords: knowledge discovery; data mining; information model Abstract. In the paper, we considered the problem to model the text document for knowledge discovery. It is urgent problem to find valuable information knowledge in the mass text documents. Data mining based on eigenvector is widely used to mine association between information of mass documents as much as possible for knowledge discovery. However, traditional information model methods for text are hard to achieve the feature vector for some special knowledge discovery task. This paper proposed a new feature selection method of text for knowledge discovery, which is useful to find the valuable words of the text for the special knowledge discovery. In addition, the dimension of the feature vector has been reduced by proposed method, which is of great help to improve the efficiency of data mining effectively.