摘要
在对大量银行机构名分析之后,采用分支限界策略[3],来得到这个将机构名定位的寻路算法(PFA,Pathfinding algorithm)。基本思路是:将中文金融机构名分成银行名、省名、市名和尾部混合名四部分,逐层剥离,去类型关键字,然后分别匹配来获取一系列原始匹配矩阵系数,再根据分支限界思想逐步修正匹配矩阵,最终得到最优解。该应用不使用数据库,纯程序语言处理,匹配精确高。
After analyzing numerous names,the paper adopts Branch and Bound StrategyC33 to obtain the Pathfinding algorithm of defining institution name. The basic idea is as following： 1. Dividing the complete institution name into hank name,province name,city name and the rear name. 2. Extracting each name as above in order. 3. Removing all type keywords. 4. Matching each part to get some raw matrix coefficients. 5. Adjusting these matrix coefficients step by step according to Branch and Bound Strategy to get the optimal solution. The application does not involve database. Data is processed by programme language completely. Matching degree is high.
出处
《计算技术与自动化》
2011年第2期83-88,共6页
Computing Technology and Automation
基金
广东发展银行长沙分行代付保费业务系统开发项目资助课题(XQ2011011306)
关键词
机构名匹配
分段匹配
组织机构名称识别
寻路算法
分支限界策略
matching financial institution names
segment matching
distinguish of organization names
pathfinding algo rithm
branch and bound strategy
