针对条件函数依赖(CFDs)对不一致数据检测不完备问题,提出基于最大依赖集(MDS)的依赖提升算法(DLA),通过获取依赖中包含的隐性依赖(RCFDs)对数据集中的不一致数据进行检测.利用动态值域调整,设置数值变化的前移和后移指针,改进原算法的枚举过程,提高了算法对连续属性的适用性,给出动态值域调整和依赖提升算法的算法流程和伪代码,并对算法的收敛性和时间复杂度进行分析.最后通过对照实验,对比了依赖提升算法和基于CFDs的检测方法的检测精度和时间代价,验证了算法的有效性.%For the incomplete detection of inconsistent data by CFDs, this paper proposes a Dependency Lifting Algo-rithm(DLA)based on Maximum Dependency Set(MDS), which detects inconsistent data in data set by acquiring Reces-sive Conditional Functional Dependencies(RCFDs)in CFDs. Presenting the dynamic domain adjustment, setting forward and backward pointers of numerical change to improve the enumeration process in original algorithm, the applicability of the algorithm to the continuous attributes is raised too. Then, this paper provides the algorithm flow and pseudo code of dynamic domain adjustment and the DLA, analyses the convergence and time complexity of them. Finally, the validity of the DLA is verified by comparing the detection accuracy and time-cost.
展开▼
