大规模混合网络数据库模糊查询算法改进仿真

Improved Simulation of Large-Scale Hybrid Network Database Fuzzy Query Algorithm

由于现有的数据库模糊查询方法不能满足用户对数据库模糊查询查全率、查准率等方面的需求,为了满足用户的这些需求,提出了一种基于数据聚类的数据库模糊查询方法。利用小波分解与小波重构相结合的方式将混合网络数据库信号分解成低频、高频两种形式,选取适当的阈值处理方法,滤除大规模混合网络数据库中数据包含的噪声。将小波系数中绝对值低于阈值的全部元素置零,其它系数保持不变,采用冗余滤除方法滤除数据库中的无意义信息,对去除冗余后的数据做聚类处理,将其作为模糊查询的样本,通过模糊C均值算法实现对数据库的模糊查询。实验结果表明,改进的数据查询方法提高了数据库模糊查询的查全率和查准率、降低了数据库模糊查询的响应时间。

Current database fuzzy query method cannot satisfy user requirement for recall ratio and precision ratio. Therefore, this research presents a new query method based on data clustering. Firstly, the research integrated wavelet decomposition with wavelet reconstruction to decompose mixing network database signal to low frequency and high frequency forms, then selected appropriate threshold processing method and filtered noise contained by data in large-scale mixing network database. The research reset all elements whose absolute values of wavelet coefficient were inferior to threshold value while other coefficients remained unchanged and used redundant filter method to filter insignificance information in database. Moreover, the research carried out clustering for the redundancy-eliminated data and used the data as sample of fuzzy query. Finally, we achieved the fuzzy query via fuzzy C means algorithm. Simulation results show that the modified data query method improves recall ratio and precision ratio. It reduced response time of the fuzzy query.