Mining Quantitative Association Rules Based On Fuzzy Set Theory

Data mining currently is the research frontier within the information science field. It had success applications in many areas. It can find the potential knowledge which hides behind the large data to forecast the trend of things development.Association rule mining is one of the key points in the research field of data mining. To deal with the problem of sharp boundary in mining quantitative association rules, researchers have introduced fuzzy set theory to data mining. Such kind of association rules is called fuzzy association rules. Using fuzzy set concept the discoved rules are more understandable to a human. However, there are some problems in mining model, including unreasonable support definition and subjectively determining membership function. Conceretly, the main work includes:1. Firstly, based on the novel support measure and Apriori algorithm, a new algorithm for mining fuzzy association rules is proposed. Experimental results illustrate the algorithm is more effective.2. Secondly, a method of determining the membership function of sample data based on the fuzzy c-means is presented. The method overcomes the defect of determining membership functions subjectively. The method is more practical especially when it is difficult to know a priori which fuzzy set will be the most suitable.3. Thirdly, examine the relationships between mined result and the minimum support value when using different fuzzy t-norms, and analysis the influence of different t-norms on the algorithm's performance.