基于粒计算的数据分片算法的问题发现

Problem-Finding of Data Fragment Algorithm Based on Granular Computing

基于粒计算的数据分片算法在考虑数据本地化的基础上,利用相似粒来划分全集数据。通过分析粒化算法的合并规则,发现在分布式数据库(Distributed Database,DDB)中,如果不考虑问题数据的出现,将会导致粒化算法产生的数据分片结果不可用,所以必须要消除问题数据。文中提出了利用黄金分割的知识进行校验,降低问题数据的产生,具有一定的可行性。文中指出如果无法保证数据的准确性,就无法使用该算法,并详细分析该算法的实际效用,对该算法的应用做了一定的限制。

Considering data localization,the data fragment algorithm based upon granular computing uses similar granular to divide the global data.After analyzing the combination rule of the granulation algorithm,data fragment result produced by the granulation algorithm is unusable if the problem data is not considered in the distributed database.Therefore,the problem data should be eliminated.In this paper,golden section is used for detection to reduce problem data which is feasible.Last,this algorithm can not be used if data is not precise,and then practical effectiveness of it is analyzed to limit its application.