The Research And Application Of Weight Fuzzy Association Rules Mining Algorithm

As information technology, particularly the rapid development of Internet technology, e-commerce came into being and gradually spread. E-commerce system database has accumulated huge amounts of data, but valuable knowledge and business decision-making is very scarce. As the main mode of data mining, association rules is used to determine the links between domains or attributes of different data sets, to find valuable dependencies between multiple domains, to recommend related products to users, to improve business profits in online shopping field. Association rules has become the most mature, most active and important research focus in data mining as its simple form and easy to interpret and understand. This article studys data mining association rules mining algorithm, and research results is used to a personalized recommendation system in a certain e-commerce sites. Main contents include the following:Firstly, sum up the current research situation, and give data mining concepts and basic methods and difficulties. Introduce focused the classic Apriori algorithm for association rule mining algorithm.Then, propose a matrix-based association rule mining algorithm (BOMA). The algorithm constructs a matrix using the orderly nature of sets, gets frequent 1-itemsets and maximal frequent sets by logical computation, and then obtains frequent K-itemsets based on gotten frequent itemsets and matrix. Improve the efficiency of the algorithm, is fast and effective on mining association rules from a large number of transaction sets.Then, for the rules lost mark and the border demarcation issue in quantitative association rule mining, a Weighted Fuzzy Association Rules Mining (NFWARM) is proposed. The algorithm transitions numerical values into fuzzy values based on the fuzzy regions and associated membership functions defined by experts, avoiding the boundary mark problems arising from the division of the interval. Meanwhile, describes the contribution of elements for the rules with attribute weights to ensure that the downward closed property of frequent itemsets without causing loss of the case.Finally, the proposed NFWARM algorithm is applied to a personalized recommendation system of an e-commerce site. Improving the user's operation friendliness, the mining results of the system can play a positive role in guiding the business decision-making.