电信欺诈预警的多粒度免疫方法  

An immunological theory inspired approach to detect telecommunications frauds

作　　者：张飞[1,2] 梁意文[1] 董红斌[1] 

机构地区：[1]武汉大学计算机学院,湖北武汉430072 [2]武警指挥学院南京分院,江苏南京210038

出　　处：《哈尔滨工程大学学报》2006年第B07期223-227,共5页Journal of Harbin Engineering University

摘　　要：电信欺诈是电信业普遍存在的一个问题，而电信欺诈检测技术、手段的滞后性、不准确性，且误报率高．人工免疫系统借鉴了机体免疫的原理，具有动态平衡性，自治性，分布性，多层性，适应性等优点，正好弥补了传统检测方法的不足．本文详细介绍了基于免疫方法的欺诈检测的方案，针对电信行业的海量数据，提出了粒度变换的概念，在不同粒度上进行建模和检测．The problem of telecommunications frauds is becoming more and more serious. In this paper, an approach is presented to analyze and detect telecommunications frauds based on the artificial immune system （AIS）. Many advantages of the organism immunology theory have been used for reference of AIS, such as dynamic balance, autonomy, multi-layer distributing and adaptability, which have just remedied the deficiency of traditional detection method. In respect of the large-scale data in the telecommunication industry, the granularity transformation, involving with modeling and measure on different grain sizes, are put forward in this paper.

关 键 词：人工免疫 欺诈检测 电信欺诈 数据挖掘 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]