Abstract: The initial focal point has a great influence on the clustering effects of traditional K-means algorithm, which makes cluster into a local optimal solution. In view of the existing problem，The algorithm that generates the initial cluster centers is proposed ,through introducing the density and nearest-neighbor idea, and these selected centers are used in K-means algorithm, getting the better text clustering algorithm called DN-K-means. The experiments results confirmed that the algorithm can produce clustering result with high and steady clustering quality.

Key words: text clustering, density, nearest neighbor, F-measure

摘要： 初始中心点的选择对于传统的K-means算法聚类效果影响较大，容易使聚类陷入局部最优解。针对这个问题，引入密度和最近邻思想，提出了生成初始聚类中心的算法，将所选聚类中心用于K-means算法，得到了更好的应用于文本聚类的DN-K-means算法。实验结果表明，该算法可以生成聚类质量较高并且稳定性较好的结果。

关键词: 文本聚类, 密度, 最近邻, F度量