Multidimensional Quantitative Association Rule Mining's Application In Food Detection

Food safety relates to human health and social stabilization. But in today's China,because many food detector mistakenly think that detection effect is direct proportionto detection times, the task of the times of detection is emphasized unilaterally. Withthe increase of the times of detection, it results to the waste of our nation's wealthresource and government's human resource. There may be some useful informationbetween records in food detection and food operator database. It can help detectiondepartment to constitute reasonable detection plan by employing data miningtechnology to find the useful information.In this thesis we describe the state of art of data mining technology, and we studydeeply one of the most important data mining tasks, i.e., mining of association rules.In the study of mining association rules, we focus on the mining of multidimensionalquantitative association rules. After analyzing the basis approaches, we propose anovel algorithm named Multi-Apriori which combines data cube with Apriorieficiently using a FDPI-tree. Multi-Apriori can discover both inter-dimensionassociation rules and hybrid-dimension association rules simultaneously, and hasperformance on the whole. Then use gridding algorithm to cluster the association rulewith multidimensional numeric attributes. The large number of association rules arereduced and the results of data mining are much easier to be interpreted.