Research On Association Rules Mining Based On Interest Measure And Genetic Algorithm

With the development of the database systems, humans have accumulated large amounts of data full of a wealth of knowledge which has guidance on human activities. These data are to be mined by people. Thereby data mining as a discipline becomes a great concern in the whole society.Association rule mining is one of the most active research methods in data mining. This paper provides a general presentation of the basic definition of association rule mining and its most basic algorithm called apriori algorithm. Then the paper takes close looks at an important method in data mining named genetic algorithm, from the aspects of the algorithm’s basic steps, basic elements, and its theoretical foundation.Next, this paper studies the traditional association rule mining based on support measure and confidence measure and puts forwards the limitations of the mining. On the basis of correlation measure the paper builds up a new interest model, and further advances a new algorithm of mining association rules grounded on the interest model. Experimental results have proved that this algorithm can avoid useless and misleading rules and generate some interesting negative association rules.Finally, within the framework of the mentioned genetic algorithm, this paper goes on with the study on combining the genetic algorithm with association rule mining. Grounded in the established interest model the paper proposes another algorithm of mining association rules based on interest measure and genetic algorithm. The effectiveness of this algorithm has been verified by experiments.