适合稀少空间特征的同位模式挖掘算法

Co-location pattern mining algorithm with rare spatial features

空间同位模式挖掘是空间数据挖掘的一个重要方向.现有的同位模式挖掘着重研究每个空间特征的平等参与,而这在具有稀少特征的空间数据中会遗漏一些很有价值的模式.首先提出了一种新的称为maxPR的度量方法,每个确定的同位规则都与具有高的maxPR值的同位模式相关.其次,阐述了maxPR的弱单调性,并利用maxPR的弱单调性提出了一种适合稀少空间特征的同位模式挖掘的有效算法.

Mining technology of spatial co-location pattern is an important direction in spatial data mining. The existing studies on co-location pattern mining emphasize on the characteristic of equal participation for each spatial feature. Some valuable patterns will be omitted in these rare spatial features. Firstly, a new measure called maxPR is proposed. Each certain co-location rule is interrelated with a co-location pattern with a high maxPR. Secondly, the weak maxPR monotonous is discussed. This property can be used to develop efficient algorithm to mine co-location pattern with high maxPR from spatial data sets with rare spatial features.