| 摘要: |
| 针对网络信息审计系统的需要,提出一种新颖的基于Episode距离的快速多模式相似串匹配算法.该算法把模式串集合转换为多个有限自动机,然后利用模式串集合建立一个状态驱动器.依次用待匹配串的字符驱动状态驱动器,由状态驱动器驱动各个有限自动机,实现了中英文混合的允许插入错误的相似多模式匹配.该算法不需要匹配每个字符,能充分利用匹配过程中本次匹配不成功的信息并结合改进的文本窗机制,跳过尽可能多的字符;能够控制每个模式串的允许错误上限;匹配速度与允许插入的错误字符教k无关.该算法在信息审计、数据库、信息检索等领域有 |
| 关键词: 信息审计 允许错误的匹配 多模式相似匹配 有限自动机 |
| DOI： |
| 分类号: |
| 基金项目:Supported bythe National Security Fundfromthe Ministry of Information Industry of China under Grant No.2001-1-010(国家信息产业部计算机网络与信息安全基金) |
|
| A Multiple Approximate String Matching Algorithm of Network Information Audit System |
|
GAO Peng,ZHANG De-Yun,SUN Qin-Dong,ZHAI Ya-Hui,LU Wu-Chun
|
| Abstract: |
| This paper shows a simple, efficient, and practical algorithm for locating all occurrences of a finite number of a finite number of keywords in a char/Chinesw character string allowing k chars inserting errors.The algorithm consists of constructing muleiple finite state single-pattern matching machines form keywords and a state-driver appled to drive all finite state finite state single-pattern matching machines,and then using the state-driver to process the text string in a single pass.Speed of the matching is independentof the amount of the inserting errors.Generally,the text string in a not need to inspect every character of the string.They skip as many characters as possible by making full use of the information in matching failure and text window mechanism.This algorithm can be widely applied to network infomation auditing,database,information retrieval,and etc. |
| Key words: information audit matching allowing errors multiple approximate string match finite state machine |
