GEOARM: an Interoperable Framework to Improve Geographic Data Preprocessing and Spatial Association Rule Mining

Geographic data preprocessing is the most expensive and effort consuming step in the knowledge discovery process, but has received little attention in the literature. For the data mining step, especially for association rule mining, many different algorithms have been proposed. Their main drawback, however, is the huge amount of generated rules, most of which are well known patterns. This paper presents an interoperable framework to reduce both the number of spatial joins in geographic data preprocessing and the number of spatial association rules. Experiments showed a considerable time reduction in geographic data preprocessing and association rule mining, with a very significant reduction of the total number of rules.