基于非参数估计与随机模拟的不确定数据流相似性度量方法

摘要

To solve the problem that the current uncertain data stream is difficult to measure the similarity, this paper proposes a method combining non-parametric estimation with stochastic simulation. The method used the non-parametric estimation to model the uncertain data stream objects, and then used stochastic simulation to calculate the error similarity between objects, judged the similarity by relative distance and absolute distance. Simulation experiment verified this method can not only measure the similarity between the uncertain objects accurately, but also can obtain fast and stable results when the object scale is large.%针对不确定数据流对象难于度量相似性的问题,本文提出一种非参数估计与随机模拟相结合的方法.本方法利用非参数估计对不确定数据流对象建模,然后利用随机模拟计算对象间的误差相似性,通过相对距离与绝对距离判断相似度.仿真实验验证了本方法不仅可以准确地度量不确定对象间的相似性,而且在对象规模较大的情况下,依然可以获得较快速和稳定的计算结果.