改进量子粒子群算法优化神经网络的数据库重复记录检测  

DATABASE DUPLICATE RECORDS DETECTION USING NEURAL NETWORK OPTIMISED BY IQPSO

作　　者：陈芬[1] 

机构地区：[1]宿迁学院计算机科学系,江苏宿迁223800

出　　处：《计算机应用与软件》2014年第3期20-21,115,共3页Computer Applications and Software

摘　　要：为了提高数据库重复记录检测效果,提出一种改进量子粒子群优化算法(IQPSO)优化BP神经网络的数据库相似重复记录检测模型(IQPSO-BPNN)。首先计算记录字段间的相似度,组成特征向量;然后采用IQPSO算法优化BP神经网络进行学习,建立最优相似重复记录检测模型,最后通过仿真实验对IPSO-BPNN的性能进行测试。实验结果表明,IQPSO-BPNN大幅度减少了数据库重复记录检测时间,提高了数据库重复记录检测精度。To improve detection effect of duplicate records in database, we propose a database similar duplicate records detection model IQPSO-BPNN, which uses the improved quantum particle swam optimisation （IQPSO） to optimise BP neural network. First, it calculates the similarity of the corresponding fields between two records and forms the eigenvectors ; and then it uses IQPSO to optimise BP neural network to learn, and builds the optimal similar duplicate records detection model; finally, through simulation experiments it tests the performance of IQPSO-BPNN. Experimental results show that the proposed model significantly reduces the detection time and improves the detection accuracy of database duplicate records.

关 键 词：数据库 重复记录 神经网络 量子粒子群优化算法 时变参数 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]