提出一种基于BP神经网络的二步检查法实体匹配新算法,将基于学习的思想引入到异构数据库实体匹配领域中,避开了传统方法计算属性权重的问题。实验结果显示,该算法很有效,能明显提高实体匹配的查准率,有较强的环境动态适应性,可以实现实体匹配的自动化。
A methodology for entities matching algorithm based on two-phase-check BP neural network is proposed. Will study the thought will introduce in the field of heterogeneous database entities matching, avoided computing the attributes weights. The experimental result show, the algorithm is very effective, and improve the precision ratio, has the stronger environment dynamic compatibility, can realize the entities matching automation.