Association Rule Incremental Updating Algorithm And Its Application In Cloud

Since our human go into the information society, the information technology has obtained a rapid development. At the same time, the data size of human's activity in our society is increasing everyday. However, how to deal with such complex, numerous and complicated data and then change them into a kind of format which our human can be easy to accept and understand is a problem for the researcher of the time.Firstly, this paper introduced the backgrounds of data mining and the main relevant research content, then makes a detailed analysis from the process of data mining and leads to the important data mining branch, association rules.During the process of theoretical study on the association rules, we firstly give a brief introduction about the classic apriori algorithm, then proposed how to efficiently mine association rules and update increment under the environment of big data, on the basis of the FUP algorithm, a new MFUP algorithm which based on association rules and incremental updating of matrix is proposed. This proposed method reduces the scan times of datasets by transforming the dataset to Boolean matrix, and the storage space is decreased by the Boolean matrix. The experiment about incremental updating of frequent items verified that the consumed time of the MFUP algorithm is less time than that of the FUP algorithm when mining association rules and updating increments of the same amount of data. In addition, with the increasing of datasets, the growth rate of consumed time is slower in the MFUP algorithm. This proposed method reduces the scan times of datasets by transforming the dataset to Boolean matrix, and the storage space is decreased by the Boolean matrix.Then combine with the MFUP and Hadoop platform, proposed a new CMFUP algorithm within cloud environment, the experiment about incremental updating of frequent items verified that the consumed time of the CMFUP algorithm is less time than that of the MFUP algorithm when mining association rules and updating increments of the same amount of data, In addition, with the increasing of datasets, the growth rate of consumed time is slower in the CMFUP algorithm, at the same time, the CMFUP algorithm can process more data than MFUP.Finally, in view of the CMFUP algorithm, a new recommended system which is based on CMFUP is put forward and designed. In order to generate the most personalized recommendation by the association rules, the paper present a new step which need to clustering the origin dataset before mining, in the end, import the true dataset into the system and show a demonstration.