集成元规则和FP-增长树方法的多层空间关联规则挖掘算法  

Multi-level Spatial Association Rules Mining Algorithm Integated Meta-rule and Frequent Pattern Tree Algorithm

作　　者：万鲁河[1] 刘万宇[1] 崔金香[1] 

机构地区：[1]哈尔滨师范大学地理系

出　　处：《小型微型计算机系统》2008年第4期715-719,共5页Journal of Chinese Computer Systems

基　　金：国家自然科学基金项目(40771154)资助;黑龙江省自然科学基金项目(TK2005-17)资助;黑龙江省高校教师骨干计划项目(151G021)资助

摘　　要：在分析典型的空间关联规则算法的基础上,针对规则冗余和挖掘伸缩性差的不足,提出基于元规则和FP增长树的空间关联规则挖掘算法,该算法不用产生候选集合,而使用最不频繁的项后缀,减少了扫描数据库的次数,大大降低了搜索开销;同时,元规则的约束,提供了好的选择性,减少了规则的冗余.本文最后以水土流失的空间要素的关联关系为例,验证算法的有效性,并与典型算法比较,本文提出的算法时间性能和空间伸缩性均优于典型算法.On the basis of analyzing the typical spatial association rules algorithm, the paper aims at the shotage of the regular redundancy and the excavation elasticity proposes a multi-level spatial association rules mining algorithm based on recta-rule and FP-Tree growth algorithm. This algorithm does need not to have the candidate set, but uses the most infrequent suffix, the reduces times of scanning database and the cost of searching greatly. At the same time, a Meta-rule restraint offers the good alternative, and reduces the rule redundancy. This article finally takes the incidence relation of special data of soil erosion as an example; verifies the algorithm, and compares with the typical algorithm. The algorithm times performance and space elasticity are proposed in this paper are all superior to the typical algorithm.

关 键 词：空间关联规则 FP-增长树方法 元规则 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]