Evaluation methods for association rules in spatial knowlegde base
Association rule is an important model in data mining. It describes the relationship between predicates in transactions, makes the expression of knowledge hidden in data more specific and clear. While the developing and applying of remote sensing technology and automatic data collection tools in recent decades, tremendous amounts of spatial and non-spatial data have been collected and stored in large spatial database, so association rules mining from spatial database becomes a significant research area with extensive applications. How to find effective, reliable and interesting association rules from vast information for helping people analyze and make decision has become a significant issue. Evaluation methods measure spatial association rules with evaluation criteria. On the basis of analyzing the existing evaluation criteria, this paper improved the novelty evaluation method, built a spatial knowledge base, and proposed a new evaluation process based on the support-confidence evaluation system. Finally, the feasibility of the new evaluation process was validated by an experiment with real-world geographical spatial data.