摘要： 提出一种二次挖掘辨别方法，以辨别系统边界外实体。通过扩展链接分析技术，令关联顶点为边的属性并定义边相似度，在此基础上，设计快速辨识算法。根据IM用户倾向于与固定社区通信的特点，挖掘社区并将社区收缩为一个点，调用快速辨识算法以提高精确度。实验结果表明，该方法是有效可行的。

关键词: 数据挖掘, 社区挖掘, 链接挖掘, 实体辨别, 即时消息网络

Abstract: This paper presents a secondary mining algorithm for entity recognition. It extends link mining techniques, using linked vertex as features of edges and definite edge similarity. A fast entity recognition algorithm can be designed. As IM user tends to communicate with fixed community, community is mined and contracted to vertex. The fast entity recognition algorithm is used to improve accuracy. Experimental results show it is effective and feasible.

Key words: data mining, community mining, link mining, entity recognition, IM network

中图分类号: 

• N945