Abstract: Fuzzy location of instances can be applied to many areas, such as biomedical image databases, geographic information system (GIS) and more. This paper investigates the spatial co-location patterns mining problem for fuzzy location of instances. Firstly, it defines the related concepts of co-location patterns mining for fuzzy location of instances, including fuzzy location of instances, location participation ratio, etc. Secondly, it proposes a basic algorithm to mine co-location patterns from fuzzy location of instances. Then, it puts forward two kinds of the improved algorithms, grid-based distance calculation and the pruning candidate patterns, so as to improve the mining performance and accelerate the co-location rule generation. Finally, by extensive experiments, this paper verifies the efficiency and effectiveness of the algorithms.

Key words:  spatial data mining, co-location patterns, fuzzy location of instances, location participation ratio

摘要： 实例位置模糊在许多领域里都有着非常重要的应用，比如生物医学图像数据库和地理信息系统（geographic information system，GIS）。研究了实例位置模糊的空间co-location模式挖掘问题。定义了实例位置模糊的空间co-location模式挖掘的相关概念，包括实例位置模糊、位置参与率等；给出了基本算法来挖掘实例位置模糊的co-location模式；提出了两种改进算法，即基于网格的距离计算和减枝候选模式，以提高挖掘性能，加快co-location规则的产生。通过大量的实验，说明了基本算法及其改进算法的效果和效率。

关键词: 空间数据挖掘, co-location模式, 实例位置模糊, 位置参与率