空间高效用Co-location模式挖掘技术初探

Primary Exploration for Mining Spatial High Utility Co-location Patterns

空间Co-location模式是一个空间特征集,集合中各个特征的实例在邻近区域内频繁并发出现.在经典的频繁模式挖掘研究中,最近的突破之一是将效用概念作为新的兴趣度量,它允许事务中同一个项可以有多个实例出现且不同的项可以具有不同价值.本文将效用概念引入到空间Co-location模式挖掘中,定义了模式效用、模式效用率等概念,提出一种基础算法挖掘空间高效用Co-location模式.接着定义了扩展模式效用,并根据它的反单调性提出一种剪枝策略:完全剪枝算法,加快了空间高效用Co-location模式的产生.最后通过大量实验来说明完全剪枝算法的效果和效率.

A spatial co-location pattern is a group of spatial features, whose instances frequently appear in the same region. In frequent pattern mining, a recent effort has been to consider utility as a new measure of interest, this concept allows that multiple instances of an item appear in the same transaction and different items have different value. In this paper,we first incorporate utility into the spatial pattern mining, define pattern utility, pattern utility rate and so on, and propose a based algorithm for mining spatial high utility co-lo- cation patterns. Then we define a new concept：Extended Pattern Utility, and propose a pruning strategy：Complete Pruning Algorithm （ CPA ）, which improve the mining performance and accelerate the spatial high utility co-location pattern generation. Finally, substantial experiments show that CPA is effective and efficient.