Novel Algorithm of Spatiotemporal Association Rules Mining Based on Event-cov- erage

In order to eliminate data redundancy of spatiotemporal database, and flexibly create spatiotemporal association patterns, and fast discover spatiotemporal association rules, firstly, this paper adopts event-coverage to create spatiotemporal mining database; the method can divide the spatiotemporal domain into some spatiotemporal transaction cells, where each cell is made of attribute values and spatiotemporal predicate values created by the concept generalization method. Then we propose a novel algorithm of spatiotemporal association rules mining based on event-coverage, which can make each spatiotemporal association pattern be mapped to a mixed radix numeral, and uses power set to compute the support. The algorithm adopts simple data structure to discover frequent spatiotemporal association patterns, it only needs to read the database once, and need not generate candidate for mining spatiotemporal association patterns. Because of them, the algorithm overcomes these disadvantages of traditional classical algorithms for discovering frequent patterns. Finally, we discuss the optimal application environments of the algorithm to mine spatiotemporal association rules. For discovering frequent spatiotemporal association patterns on the application environments, these experimental results indicate that the algorithm is better than these traditional classical mining frameworks, particularly, the Apriori framework and the FP-growth framework.