文档聚类中k-means算法的一种改进算法  

An Improved k-means Algorithm for Documents Clustering

作　　者：万小军[1] 杨建武[1] 陈晓鸥[1] 

机构地区：[1]北京大学计算机研究所文字信息处理技术国家重点实验室,北京100871

出　　处：《计算机工程》2003年第2期102-103,157,共3页Computer Engineering

摘　　要：介绍了文档聚类中基于划分的k-means算法，k-means算法适合于海量文档集的处理，但它对孤立点很敏感。为此，文章提出将聚类均值点与聚类种子相分离的思想，并具体给出了基于该思想的对k-means算法的改进算法。实验表明，该改进算法比原k-means算法具有更高的准确性和稳定性。This paper first introduces the partitioning-based k-means algorithms for documents clustering. The k-means algorithm adapts to processing the vast amount of documents, but it is sensitive to outliers. So this paper puts forward an idea to separate the clustering centroid from the clustering seed and brings forward an algorithm based on this idea to improve the k-means algorithm. The paper shows the results of the experiments to prove that this algorithm is more veracious and stable than the k-means algorighm.

关 键 词：文档聚类 K-MEANS算法 划分聚类算法 数据库 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论]