中文摘要：

针对互联网舆情管控领域信息量大，时效性强，往往偏重于某些方向，如社会热点、焦点，或反动、黄色言论等的特点，文中把基于密度的聚类思想引入传统K-Means算法，提出全新的DK聚类算法，并且基于DK算法构建中文文本聚类模型，重点对互联网媒体发布信息进行主动热点发现研究。用实验验证中文聚类模型的具体性能，证实了该模型的有效性和实用性。

英文摘要：

In the information booming era, Intemet informtion control and supervision always need to deal with numerous update information and focusc on some specific areas such as social focus, hot topics, anti - social statement and pomo information. Considering all these features, create a Chinese text clustering model and specialized in Interact information hotspots discovery on initiative. It proposes the density based DK solution also combined the strength of K - Means algorithm and the feasibility is justified in the experiment.