Research On Incremental Updating Association Rules Mining Based On Apriori Algorithm

Recently, information technology has been gradually applied to all aspects of social production and service, with massive and endlessly growing of data. How to dig out the hidden knowledge from the massive, and especially dynamically changed data, has currently become the focus research, which can not only help to make decision, but improve the generated of data and information technology perfect. Because of it, more and more researchers conducted deeply research, and make Association Rules Mining an important and active research direction. All of this achievement laid a solid foundation for this filed. Concerning the shortcomings in Association Rules Mining, this paper, according to the change of data set and user demand, starts from the basic algorithm, and makes a research for the Incremental Updating Association Rules.First, analyzes the defects of Apriori algorithm, and then makes an optimization of Apriori using the idea of set operations and get a new algorithm:TSApriori. In this algorithm, the frequent itemsets and its corresponding transactions are all represented by set, which provide a foundation to the dynamic update of association rules. After that, it constructs the frequent itemsets tree to optimize the generation of association rules, and introduces the concept of correlation degree to verify the candidate rules to ensure the accuracy of the generated rules. Secondly, based on TsApriori, it combined the thoughts with FUP and IUA organically, and Proposed a new algorithm IUTS, to adapt the update of the dataset and the change of minimum support at the same time. As a result, the administrative and criminal case data of he Dalian Municipal Public Security Systems are used to verify these algorithm.The TsApriori represented, only needs to scan the database once, and optimizes the connection operations, avoid to generate candidate K itemsets, which improves the efficiency of mining frequent itemsets. Use the frequent itemsets tree to optimize the generation of association rules, can reduce the redundancy and false rules, improved the efficiency and accuracy of the algorithm. The IUTS, using the existing results, avoid scanning the database repeatedly. These ensured the accuracy and also have high efficiency and good extensible.