Study Of Maintenance Algorithm For Association Rule

The mining of association rules is an important research field in data mining and is of great value in application. The discovery of the frequent item sets is a key to the mining of association rules. The maintenance of the association rules and the frequent item sets is a nessesary research content for the database varies.Based on analysis and study on presented algorithms, An association rules algorithms for solving some relative problems in association rules data mining is proposed.The main contents are as follows:1. An algorithm solving frequent itemset based on item increasing is proposed. The algorithm will get all frequent itemsets by scanning database only once, and this will boost up pertinence and validity of the candidate items and will improve the efficiency of candidate items supporting item counting. During the procedure of resolving the maximal frequent itemsets, a large amount of middle results are not needed, therefore memory space will be saved. By means of comparison analysis, this method improves the efficiency and performance of the mining.2. This algorithm makes use of the results of mining and will get frequent itemsets of the item updated data set by scanning the newly-added data set only once without rescaning the original one, it greatly reduces the run-time and improves the mining efficiency.3. Puts forward an updating algorithm for association rules in which the size of data sets is reduced is proposed with the supporting and confidence limits unchanged.4. The dissertation presents a fast focus and meaningful updating algorithm for association rules, with the item data sets unchanged, but the minimum supporting limit changed is presented. It explores how to make the best use of the known information of mining to avoid rescanning data set.5. A prototype system is implemented for above algorithms.The experiments show the performance.