M.-S. Hacid et al. (Eds.): ISMIS 2005, LNAI 3488, pp. 342 – 353, 2005. © Springer-Verlag Berlin Heidelberg 2005 Mining and Filtering Multi-level Spatial Association Rules with ARES Annalisa Appice, Margherita Berardi, Michelangelo Ceci, and Donato Malerba Dipartimento di Informatica, Università degli Studi, via Orabona, 4 - 70126 Bari - Italy {appice, berardi, ceci, malerba}@di.uniba.it Abstract. In spatial data mining, a common task is the discovery of spatial association rules from spatial databases. We propose a distributed system, named ARES that takes advantage of the use of a multi-relational approach to mine spatial association rules. It supports spatial database coupling and discovery of multi-level spatial association rules as a means for spatial data exploration. We also present some criteria to bias the search and to filter the discovered rules according to user’s expectations. Finally, we show the applicability of our proposal to two different real world domains, namely, document image processing and geo-referenced analysis of census data. 1 Introduction Spatial data mining investigates the problem of extracting pieces of knowledge from data describing spatial objects, which are characterized by a geometrical representation (e.g. point, line, and region in a 2D context) and a position with respect to some reference system. The relative positioning of spatial objects defines implicitly spatial relations of different nature, such as directional and topological. The goal of spatial data mining methods is to extract spatial patterns, that is, patterns involving spatial relations between mined objects such that they are certain, previously unknown, and potentially useful for the specific application [10]. In [13] the authors have proposed a spatial data mining method, named SPADA (Spatial Pattern Discovery Algorithm), that discovers spatial association rules, that is, association rules involving spatial objects and relations. It is based on an Inductive Logic Programming (ILP) approach to (multi-) relational data mining [5] and permits the extraction of multi-level spatial association rules, that is, association rules involving spatial objects at different granularity levels. For each granularity level, SPADA operates in three different phases: i) pattern generation; ii) pattern evaluation; iii) rule generation and evaluation. In this paper, we describe the integration of SPADA in a full-fledged spatial data mining system, named ARES (Association Rules Extractor from Spatial data), that assists data miners in the complex process of extracting the units of analysis from the spatial database, specifying the background knowledge on the application domain and defining some form of search bias. The last aspect is particularly relevant, since the