条件函数依赖的增量计算  

Incremental calculation of conditional functional dependencies

作　　者：刘波[1] 周健昌[1] 

机构地区：[1]暨南大学信息科学技术学院,广东广州510632

出　　处：《系统工程与电子技术》2015年第11期2640-2647,共8页Systems Engineering and Electronics

基　　金：国家自然科学基金(U1431227);广东省科技计划项目(2013B010401017);广东省自然科学基金(S2012010008831)资助课题

摘　　要：条件函数依赖是对传统函数依赖的扩展,它通过引入条件模式,使其语义比函数依赖更精确、表达能力更强。然而,条件函数依赖的计算需要消耗较多的时间,为了提高条件函数依赖挖掘的效率,研究了条件函数依赖增量维护方法。针对数据集增加、删除、修改3种情况分别分析了条件函数依赖集变化规律,提出了条件函数依赖的增量计算算法,从而能够在数据库变化情况下,高效、动态地维护条件函数依赖。同时,在理论上对算法中关键步骤的正确性进行了论证,并通过实验验证了算法的有效性。A conditional functional dependency （CFD） is an extension of the traditional functional dependen- cy （FD）. By introducing the conditional pattern, the CFD is more accurate and more expressive than FDs in se mantics. However, it is time-consuming for computing CFDs. In order to improve the efficiency of CFDs, the incremental maintenance method for CFDs is studied. The changing rules on the conditions of three different sit- uations （i. e. , dataset insertion, deletion, update） are analyzed, and an incremental algorithm for calculating CFDs is proposed, so that we can efficiently and dynamically maintain CFDs while the database is changing. At the same time, the correctness of key steps of the algorithm is demonstrated, and the validness of the algorithm is verified through the experiments.

关 键 词：增量计算 条件函数依赖 数据挖掘 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论]