大型数据库中多属性资源分配均衡技术仿真

Simulation of Multi Attribute Resource Allocation Balance in Large Database

大型数据库中对多属性资源的均衡性分配,能够有效提高数据库多属性资源利用率,加快数据库运行速度。对多属性资源的分配均衡,需要获取多属性资源特征数据的初始聚类,确定网络隐含层分配节点,完成大型数据库中多属性资源的分配均衡。传统方法对多属性资源属性进行约简,从多属性资源数据挖掘的角度进行基于属性重要度的多属性资源分配,但忽略了对网络隐含层分配节点的确认,导致多属性资源的分配均衡度较差。提出基于主题相关性挖掘的大型数据库多属性资源分配均衡优化方法。构建决策随机变量构造多属性资源主题关联模型,依据皮尔逊相关系数实现多属性资源主题间近似性的度量,提取出丰富的多属性资源特征,采用二次聚类获得多属性资源特征训练数据的初始聚类,确定神经网络隐含层分配节点,提高多属性资源均衡性分配速度。实验结果表明,所提方法有效改善了多属性资源分配平均时间,提升了分配精度,具有较好的鲁棒性。

Traditional equilibrium allocation method of multi -attribute resource in large database always ignores to confirm allocation node of hidden layer of network, which leads to poor allocation balance degree. In order to overcome the defect, based on topic relevance mining, an optimization method of the allocation balance is presented in this article. Decision random variable was built to construct the topic relevance model of multi - attribute resource. According to Pearson＇s correlation coefficient, measurement of similarity between topics of the multi - attribute resource was achieved. Abnndant features of the multi - attribute resource were extracted. Initial cluster of training data of the feature was obtained via twice clustering to confirm the allocation node and improve allocation speed of the equilibrium. Simulation results show that the method can improve mean time of the allocation and precision. It has better robustness.