集成同位模式的空间关联规则挖掘方法  

MINING METHOD OF SPATIAL ASSOCIATION RULE INTEGRATING CO-LOCATION PATTERNS

作　　者：向俊[1] 王静[2] 

机构地区：[1]广西广播电视大学教学资源与技术中心,广西南宁530022 [2]中国电信集团山东分公司,山东济南250101

出　　处：《计算机应用与软件》2013年第12期185-189,共5页Computer Applications and Software

基　　金：广西教育厅科研项目(201106LX868;201106LX864);广西广播电视大学中青年项目(2012ZQN06)

摘　　要：在空间数据挖掘中,使用传统的频繁集挖掘方法会导致空间实体重复计数,而且未考虑到空间实体之间的关系以及空间实体与周围环境的相关性将产生大量不相关的空间信息。针对以上存在的不足,提出集成空间同位模式挖掘的空间关联规则挖掘模型及算法。引入"地理学第一定律"和"生物群落构建"思想,首先对空间实体特征分布进行优化划分及连续空间分布的离散化处理;然后消除冗余的空间信息,并使用空间同位算法挖掘不同空间实体特征的隐含关系,由实体的空间属性、非空间属性及空间关系来构建空间事务数据库;最后在空间事务数据库中进行空间关联规则挖掘。实验结果表明,该模型和算法是有效的,考虑到空间实体特征间的相关性,经过冗余信息消除方法和连续空间分布的离散化处理,在构建的空间事务数据库中能发现多种有意义的空间关联规则。In spatial data mining, to use traditional frequent item-set mining method will lead to duplicated counting of spatial entities; moreover, it does not take into account that the relations between spatial entities and the correlation of spatial entities and their surrounding environments will produce a large amount of irrelevant spatial information. In order to overcome the deficiency, in the paper we propose She model and the algorithm of spatial association rule mining which integrates the spatial co-location patterns mining. The idea of ＂the first law of geography＂ and ＂the biocoenosis construction＂ are introduced. Firstly, it carries out the optimised partition on spatial entities features distri- bution and the discretised process on continuous space distribution. Secondly, it eliminates the redundant spatial information, and uses spatial co-location algorithm to mine the implicit relationships of different spatial entities features. The spatial attributes, no-spatial attributes and spatial relationships of the entities are employed to construct the spatial transaction database. Lastly, the mining of spatial association rules is conducted in spatial transaction database. Experimental results show that the model and the algorithm are effective, the correlation between the spatial entities features is considered. Many significant spatial association rules can be discovered from the constructed spatial transaction database after the elimination of the redundant information and the discretised processing on the continuous space distribution.

关 键 词：空间关联规则 生物群落原理 数据挖掘 空间同位模式 

分 类 号：TP208[自动化与计算机技术—检测技术与自动化装置]