空间拓扑关系相似性度量是空间数据相似性度量的重要组成部分, 它是空间数据检索和空间场景相似查询的基础和关键。本文旨在建立一种能够同时度量简单和复杂拓扑关系间的相似性度量模型。首先, 以9 个拓扑谓词提出一种简单拓扑关系间相似性度量方法, 即定义简单拓扑关系之间距离为两个9 交集矩阵之间的距离, 从而得到简单拓扑关系之间的相似度。然后, 综合考虑实体集合间的数量相似度和维数相似度,建立实体集合间的简单拓扑关系相似性度量模型, 并在上述实体集合间简单拓扑关系相似性度量的基础上, 采用分解-组合的策略先将复杂拓扑关系分解为若干个局部拓扑关系, 通过组合局部拓扑关系之间的相似度, 直至建立复杂拓扑关系间的相似性度量模型。实验表明, 制图综合中的选取对空间实体集合之间拓扑关系相似度的影响最大, 其他因素影响较小,因此, 其能以拓扑关系相似度来度量制图综合对拓扑关系的改变程度。

Similarity measuring of spatial topological relations is the important part of similarity measuring of spatial data, and also is the basic and key technology of spatial data retrieval and spatial scene query. Its meaning is to measure the similarity of topological relationships between multiple data entities in different sources, different sources scales of the same region. Common topological relations have been abstracted into nine topological predications. Current researches mainly focus on the topological relations similarity measuring between two simple entities, but mostly do not involve topological relations similarity measuring for the entire data sets, as well as the complex line targets. In this paper we present a method of measuring simple topological relations based on 9- intersection matrix, that is, the distance between two 9- intersection matrixes as the simple topological relations distance to measure the differences between two simple topological relations, so that we can get a simple topological relations similarity. Then considering the quantity similarity and dimension similarity between entity sets, we can get the simple topological relations similarity measuring model between entity sets. In this paper we establish a similarity measuring model of complex topological predication by using the strategy of decomposing- combination based on the simple topological relations similarity measuring model. Firstly, the complex topology relationship is broken down into a number of local topological relationships. Then through a combination of local topological relations similarity, we get the complex topology relationship similarity measuring model. At last, the method is used to measure similarity of different scales and different sources data. Experimental results show that the selection of cartographic generalization impact the topological relations similarity between entity sets mostly, and other factors with smaller impacts to the experimental data in this article. Experimental results also demonstrate that the topological relations similarity can be used to measure the changing degree of topological relations caused by the cartographic generalization.