模糊度阈值范围内模糊对象的co-location模式挖掘

【作者】 欧阳志平；

【导师】 王丽珍；

【作者基本信息】 云南大学 ， 计算机软件与理论， 2012， 硕士

【摘要】 空间co-location模式代表的是一组空间对象,它们的实例在空间中频繁的关联。空间co-location模式挖掘是空间数据挖掘的一个重要研究方向,在现实生活中有着十分广泛的应用。人们已经在确定及不确定数据上对co-location模式挖掘问题进行了大量研究并获得了很多成果,但在模糊数据上进行的研究几乎没有。模糊数据可以应用于许多领域,比如GIS和生物医学图像数据库。本文研究模糊度阈值范围内模糊对象的co-location模式挖掘问题。首先,介绍了空间co-location模式挖掘的相关概念、性质、相关工作及算法。其次,研究了模糊对象的空间co-location模式挖掘问题。提出了相关定义及算法,包括一个基本算法和四个改进算法,改进算法包括剪枝模糊对象、减少实例间连接、改进剪枝步和基于网格的距离计算,并对算法的时间性能及挖掘结果进行了实验分析。第三,针对模糊对象的空间co-location模式挖掘只能在单一模糊度阈值下进行的问题,论文对模糊度阈值范围内模糊对象的co-location模式挖掘问题进行了研究。提出了相关定义,并在此基础上给出了一个基本算法。为了提高基本算法的挖掘效率,提出了减少挖掘次数和缩小挖掘范围两个改进算法,并用实验比较了基本算法和改进算法的时间性能。第四,将模糊对象的co-location模式挖掘应用于三江并流项目中。第五,对论文的全部内容进行总结并对未来的工作做出展望。更多还原

【Abstract】 Space co-location patterns represent a group of spatial objects whose instances are frequently associated in the space. Space co-location pattern mining is an important research direction for spatial data mining, and has a very wide range of applications in real lift. The mining co-location pattern problem for certain and uncertain data had been investigated in the past, but not for fuzzy data. Fuzzy data could be applied to many areas such as GIS and biomedical image databases. This paper investigates the spatial co-location pattern mining problem for ambiguity range.First, the definitions, theorems, related work of the spatial co-location pattern mining and algorithms are introduced.Second, we investigate the spatial co-location pattern mining problem for fuzzy objects. The related concepts and algorithms of spatial co-location patterns mining on fuzzy objects are proposed. Algorithms includes FB algorithm, the pruning objects, reducing of the operation joining between spatial instances, optimizing the pruning steps and grid-based distance calculation. By extensive experiments, we analyzed the performance and results of the algorithms.Third, because the spatial co-location pattern mining for fuzzy objects only in a single probability threshold, so the paper investigates the problem for ambiguity range. The related concepts are defined, and on this basis, a basic algorithm has been proposed. To improve the mining performance, two kinds of the improved algorithms---reducing the number of mining and narrowing the excavation area are put forward. The time performance of the algorithms has been analyzed.Forth, the fuzzy object co-location pattern mining applied in a "three parallel rivers" project.At the last, conclusion and future work were presented.更多还原