Towards an Improved Approach for Extracting Spatial Association Rules: an Empirical Study in Algeria

Abstract

The increasing production of geographical charts generates large volumes of data that exceed human capacities for analysis. From where is it interesting to apply the extraction techniques of knowledge, such as Data mining, to the geographical (spatial) databases to discover models or hidden rules? Much research was conducted on the discovery of knowledge in relational databases. However, some work treats the extracted knowledge in the spatial databases as particular data because they have a spatial component describing the objects or phenomena located on the ground. In this article, we present a method for extracting Association Rules (AR) from a spatial database. The proposed approach improves ARGIS in the direction of considering several reference layers instead of one. Moreover, the rules discovered connect several layers or objects; the latter have spatial and non-spatial properties. The method was tested on a spatial database including four layers: Population, Hypsography, Surface Water, and underground water in Algeria. Where the generated rules give new information that cannot be extracted by the traditional methods of spatial analysis.