Binary Information Press
In the basis of principle of granularity, the authors aim to study the concept extraction method in granules space based on cloud model. A fast information granulation extraction algorithm is proposed after careful analysis of the existing methods. According to the characteristics of data distribution, the algorithm can partition date granules and extract concept automatically without any experience, and has less time-consuming, a high degree of polymerization. In the experiments of text feature selection in text mining, the algorithm receives a high-performance and its efficiency is fully proved.