基于相似度的离群模式发现模型
发布时间:2019-04-10 10:39
【摘要】:提出了基于相似度的离群模式发现模型 ,该模型主要利用知识属性集分析离群点 ,既能够处理离群点的数值属性 ,又能够处理其类别属性 ;通过组间相似度从中发现离群模式 ,不仅回避离群点数量少的缺陷 ,还利用了离群点的隐含语义 .给出了在银行结售汇交易数据上进行的实验分析结果 ,模型发现了某地区的 3个可疑模式 ,该结果为金融犯罪分析提供有利线索 ;利用不同子空间角色划分 ,可以发现个人、地区等不同对象间的异常资金流动 ;模式发现算法具有线性时间复杂度 ,在实际应用中具有较好的性能 .结果表明模型能检测出可疑资金流动序列 ,为反洗钱工作提供有意义的线索 .
[Abstract]:An outlier pattern discovery model based on similarity is proposed. The model is mainly used to analyze the outliers by using the knowledge attribute set, which can not only deal with the numerical properties of the outliers, but also can handle the class attributes of the outliers, and find the outliers through the similarity among the groups. It not only avoids the defects of the number of outliers, but also uses the hidden semantics of the outliers. The result of the experimental analysis on the transaction data of the bank settlement is given. The model has found three suspicious patterns in a certain area. The result provides an advantageous clue for the analysis of the financial crime, and can find the individual by using the different sub-space characters. The pattern discovery algorithm has a linear time complexity and has better performance in practical application. The results show that the model can detect the flow sequence of suspicious funds and provide meaningful clues to the work of anti-money laundering.
【作者单位】: 华中科技大学计算机科学与技术学院 华中科技大学计算机科学与技术学院
【基金】:国家“十五”重大科技专项基金资助项目 (2 0 0 1BA10 2A0 6 11) .
【分类号】:D917
本文编号:2455732
[Abstract]:An outlier pattern discovery model based on similarity is proposed. The model is mainly used to analyze the outliers by using the knowledge attribute set, which can not only deal with the numerical properties of the outliers, but also can handle the class attributes of the outliers, and find the outliers through the similarity among the groups. It not only avoids the defects of the number of outliers, but also uses the hidden semantics of the outliers. The result of the experimental analysis on the transaction data of the bank settlement is given. The model has found three suspicious patterns in a certain area. The result provides an advantageous clue for the analysis of the financial crime, and can find the individual by using the different sub-space characters. The pattern discovery algorithm has a linear time complexity and has better performance in practical application. The results show that the model can detect the flow sequence of suspicious funds and provide meaningful clues to the work of anti-money laundering.
【作者单位】: 华中科技大学计算机科学与技术学院 华中科技大学计算机科学与技术学院
【基金】:国家“十五”重大科技专项基金资助项目 (2 0 0 1BA10 2A0 6 11) .
【分类号】:D917
【相似文献】
相关重要报纸文章 前3条
1 赵士峰 ;离群众近些,再近些[N];人民公安报·交通安全周刊;2005年
2 孙莉莉 李平伟;办事公道了 离群众近了[N];法制日报;2003年
3 郭冬青;建路管路为民富 一片朝阳铺坦途[N];辽宁日报;2003年
,本文编号:2455732
本文链接:https://www.wllwen.com/shekelunwen/gongan/2455732.html