在信息处理研究领域,现有的大多数聚类算法都需要人为地给出一些参数.然而,在没有先验知识的情况下,人为地确定这些参数是十分困难的,而且现有的聚类算法的时空效率也有待于进一步提高.为了解决这一难题,首先根据样本分布特性,通过数学分析,得到确定样本空间划分间隔数的数学函数,然后,再根据样本分布特性,采用爬山的策略得到样本类的划分,最后提出了一种实用而高效的聚类算法.从多个角度分析了该算法的性能,并将该算法应用于中文文本聚类.理论分析和应用结果都表明,该算法不仅不需要人为确定参数,同时,还可以提高信息处理的时空效率和性能.

In the research on IR (information retrieval), lots of clustering algorithms have been developed, and in most of them some parameters should be determined by hand. However, it is very difficult to determine them manually without any prior domain knowledge. To solve this problem, an applicable and efficient clustering algorithm is presented. It aims at avoiding any parameter to be determined by hand, and at the same time, improving the efficiency of clustering and the property of IR. The new clustering algorithm is analyzed on several facets and applied later to cluster Chinese documents. The results of the application confirm that the new clustering algorithm is very applicable and efficient.

Supported by the National Natural Science Foundation of Chmaunder Grant No.60173027(国家自然科学基金)