【作者】 王丽珍； 周丽华； 肖清；

【Author】 Wang Lizhen,Zhou Lihua,and Xiao Qing (Department of Computer Science and Engineering,School of Information Science and Engineering,Yunnan University, Kunming 650091)

【机构】 云南大学信息学院计算机科学与工程系；

【摘要】 异常点检测是数据挖掘领域中的重要研究方向,但传统的异常点检测算法存在许多不足,比如缺乏用户的导向和控制,在挖掘之前就确定了挖掘空间,不能动态地根据中间挖掘结果选取挖掘空间等缺点.数据联机分析挖掘(OLAM)是将OLAP和数据挖掘有机结合,它能方便地对任何一部分数据或不同抽象级别的数据进行挖掘,得到更为详细、更为有用的信息,成为数据挖掘研究的热点.首先针对现有局部异常度定义复杂的问题,提出一种新的基于k距离的局部异常度定义.在此基础上讨论了基于空间数据仓库的动态异常点检测方法,它们是基于约束的异常点检测算法和基于OLAP的探测式异常点检测算法,并在澜沧江空间数据仓库上验证了新的局部异常度定义和所提算法的正确性和有效性.更多还原

【Abstract】 The outlier detection is an important research direction in data mining applications.But the traditional technology has some shortcomings,such as lacking the user’s guiding and controlling,and being incapable to select mining space dynamically according to the mid-results.Therefore,it is necessary to introduce a new method.On-line analytical mining(OLAM) is the combination of data mining(DM) and on-line analysis processing(OLAP).OLAM can conveniently mine any part of data or different abstractive level of data,to get the more detailed and more useful information.So, OLAM becomes research hotspot in data mining.In this paper,first,a new definition of local outlier factor(LOF) is put forward in order to reduce the difficulties of understanding and calculating the LOF.Second,the outlier mining methods based on data warehouses are discussed.Two algorithms are provided.One is the constraint-based exception detection approach.Another is OLAP-based dynamic exception mining.Finally,experimental results on the Lancang River spatial data warehouse verify the correctness and efficiency of the definition and algorithms proposed.更多还原