J4

• 计算机科学 • 上一篇    下一篇

基于有权重超图的离群点检测

张 强1, 李永丽2, 董立岩3, 李 威3, 张晓辉4   

  1. 1. 白城师范学院 计算机系, 吉林省 白城 137000; 2. 东北师范大学 计算机学院, 长春 130024; 3. 吉林大学 计算机科学与技术学院, 长春 130012; 4. 长春市公安消防支队, 长春 130062
  • 收稿日期:2007-01-30 修回日期:1900-01-01 出版日期:2007-07-26 发布日期:2007-07-26
  • 通讯作者: 李永丽

Outlier Testing Methods Based on Weighted Hypergraph

ZHANG Qiang1, LI Yongli2, DONG Liyan3, LI Wei3, ZHANG Xiao hui4   

  1. 1. Department of Computer, Baicheng Teachers College, Baicheng 137000, Jilin Province, China;2. School of Computer Science, Northeast Normal University, Changchun 130024, China;3. College of Computer Science and Technology, Jilin University, Changchun 130012, China;4. Changchun Public Security Bureau, Changchun 130062, China
  • Received:2007-01-30 Revised:1900-01-01 Online:2007-07-26 Published:2007-07-26
  • Contact: LI Yongli

摘要: 基于有权重支持度框架的关联规则挖掘算法和超图分割算法, 给出一种新的基于有权重超图模型的离群点检测算法WHOT(Weighted Hypergraphbased Outlier Test). WHOT算法根据有权重支持度的定义, 重新设计了基于有权重支持度框架的关联规则挖掘算法, 并挖掘出数据集中的重要关联规则, 形成超图. 在超图上应用超图分割算法, 得到聚类集合, 再结合项权重和事务权重的定义, 判断一条记录是否为离群数据.

关键词: 数据挖掘, 离群点, 超图, 权重

Abstract: The paper presents an algorithm called WHOT (Weighted Hypergraphbased Outlier Test), which is based on weighted association rule mining algorithm and hypergraph partitioning algorithm. The association rule mining algorithm was redesigned. Hypergraph was constructed by mining significant association rules in data set. Cluster set was obtained by using the hypergraph partitioning algorithm. After that we defined the measures to judge whether a vertex in hypergraph or a record in dataset is an outlier.

Key words: data mining, outlier, hypergraph, weight

中图分类号: 

  • TP18