依据信息论的思想,从文档信息量变化的角度,对文本聚类的过程进行了分析,研究了信息量在层次聚类过程中所呈现的规律性,进而提出一种基于信息量模型的最小熵差算法.并对文中所提出的算法进行了仿真试验,结果表明最小熵差算法在聚类准确率上整体优于k-means算法和最小熵增加算法.“,”Probabilistic hierarchical clustering based on document information quantity. From an information theory angle,we study latent relations between document information quantity and document classification. A minimum entropy difference text clustering algorithm is proposed based on document information quantity. Experimental results showed that the algorithm has higher accuracy than most current algorithms like k-means and minimum entropy increase.