Abstract: A semi-supervised clustering algorithm based on classifying is proposed.This algorithm classifies original data set roughly using a few labeled data，extends the method of selecting cluster center on the basis of traditional k-means clustering；then clusters data set generally with k-meansGuider method；Finally，the cluster results are integrated.The experimental results on UCI machine learning benchmark datasets show that this method can effectively improve the clustering performance.

Key words: semi-supervised learning, clustering, k-means clustering

摘要： 提出一种基于分类的半监督聚类算法。充分利用了数据集中的少量标记对象对原始数据集进行粗分类，在传统k均值算法的基础上扩展了聚类中心点的选择方法；用k-meansGuider方法对数据集进行粗聚类，在此基础上对粗聚类结果进行集成。在多个UCI标准数据集上进行实验，结果表明提出的算法能有效改善聚类质量。

关键词: 半监督学习, 聚类, k均值聚类