CAJ | 학술논문

在CStor云数据库访问中，细粒度云存储数据的信息特征存在个体较大差异，访问路径产生多频偏移，数据访问调度性能受限。传统方法中采用PSO局部搜索方法进行访问路径修正，搜索效率低，个体更新较慢。提出一种基于细粒度云存储数据迭代恢复匹配的云数据库访问路径偏移修正算法，以PSO多信息融合Pareto支配集为指导，构建多信息融合细粒度数据的主特征模型，采用迭代恢复匹配方法增加了局部搜索提高搜索到全局最优解的几率，实现路径偏移修正。仿真实验采用3组测试集对CStor云数据库进行访问调度和路径偏移修正，结果表明采用该算法，实现对访问路径的动态实时修正，收敛距离较小，提高了CStor云数据库访问和数据挖掘性能。
재CStor운수거고방문중，세립도운존저수거적신식특정존재개체교대차이，방문로경산생다빈편이，수거방문조도성능수한。전통방법중채용PSO국부수색방법진행방문로경수정，수색효솔저，개체경신교만。제출일충기우세립도운존저수거질대회복필배적운수거고방문로경편이수정산법，이PSO다신식융합Pareto지배집위지도，구건다신식융합세립도수거적주특정모형，채용질대회복필배방법증가료국부수색제고수색도전국최우해적궤솔，실현로경편이수정。방진실험채용3조측시집대CStor운수거고진행방문조도화로경편이수정，결과표명채용해산법，실현대방문로경적동태실시수정，수렴거리교소，제고료CStor운수거고방문화수거알굴성능。
In the CStor cloud database access research, there is big difference between individual information characteris-tics of fine-grained cloud storage data, the access path generates multi frequency offset, data access scheduling perfor-mance is limited. The traditional method of using PSO local search method for accessing path correction, the search efficien-cy is low, and individual updating is slow. An access path offset correction algorithm based on fine-grained cloud storage data and iterative restoration matching cloud database is proposed, PSO multi information fusion is taken, and Pareto domi-nating is set as a guidance, the model of multi information fusion in fine-grained data recovery is constructed, iterative matching method is used, which increases the probability of local search to improve the global optimal solution perfor-mance, then path offset correction is realized. The simulation experiment using three test functions as the test set for CStor cloud database accessing. Results show that the new method realizes the dynamic and real-time correction on accessing path offset, and the convergence distance is smaller, the CStor cloud database access and data mining performance are im-proved.