Abstract: Detecting and analyzing outliers are of great importance in many applications，including telecom fraud detection，disease diagnosis，and network invasion detection，etc.Moreover，many of these fields require good real-time performance.The key attribute subspace of outliers is helpfull to find out the extended knowledge of identified outliers，such as their origin and features.Having discussed the relation of outlying individual and its outlying attribute values，this paper proposes a real-time searching algorithm for key attribute subspace of outliers based on data exploration analysis mode.Experimental results show that the approach is scalable and it can efficiently satisfy the demand of real-time outlier analysis.

Key words: key attribute subspace, outlier analysis, data exploration analysis, real-time algorithm

摘要： 离群数据挖掘与分析在通信欺诈检测、疾病诊断和网络入侵检测等多个领域具有十分重要的意义。离群数据关键域子空间可以获得数据离群起源与特征等相应的延伸知识。通过对离群数据对象与其属性值的关系讨论并基于探索性数据分析方法，提出了一种离群数据关键域子空间实时快速搜索算法。实验结果表明提出的算法是有效的，可以满足大多数实时性检测与分析要求。

关键词: 关键域子空间, 离群分析, 数据探索性分析, 实时算法