Research On Association Rules Mining Of Big Data

Recent years, our society is undergoing dramatic transformation and the development of science and technology changes has already gone far exceeded our imagination, especially in the computer that is a significant step forward in artificial intelligence research. A variety of novel techniques, algorithm and the concept is continuously changed our life. The data has at an unprecedented rate of growth and accumulation. The age of big data into Nolan's arrival means that information systems evolution models more complete and mature stages. Big data is seen as the most revolutionary and important milestone in recent decades.Association rule is a very important part of data mining. The concept of data mining is usually from a large number of databases or a large number of data set to discover previously unknown relationship between the value of a certain relationship or to verify some of the relationships have been known. The traditional association rule mining algorithm is only based on the simple production of attributes in the database, it does not take into account the specific significance of whether there is a value or redundancy. For the arrival of the era of big data, how to find the association rules we previously unknown from the complex data environment is the common goal of researchers. Due to the data with time constant, the data stored in the database into exponentially, the mutual relation of data has become complicated, and that seems to have no any contact data, through the association rules mining algorithm and obtained the certain relations are more likely to be high.Through a lot of reference documentation and books, learning and evaluation algorithm of association rules mining in the past, the paper as a whole was divided into the subjective and objective aspects of association rules of data mining research specific research contents are as follows:(1) The objective mainly through the study of the current generated by the algorithm in association rules, found some association rules algorithm of association rules generating redundant association rules than the real number of valuable number of association rules, redundant rules will not only hinder the researchers analyze and understand, but also greatly reduced in overall usage of association rules. Aiming at the problem of redundant association rules, this paper presents a removal of first-order predicate expression based on business data redundant association rules method, using first-order predicate expression to represent the association rule, converted by the equivalent formula, and the algorithm and equivalent matrix converts the predicate formula of adjacency matrix, then the algorithm of redundant rules to delete. Experimental raw data for the UCI data sets and association rules algorithm for using WEKA to association rules. Removal of redundant rules by using MATLAB and Java implementation.(2) The subjective aspect is from a user's interest measure to do research in this direction, taking into account the subjective interests of optimization algorithms remove redundant Association Rules algorithm combining with the objective will complement its lack of ability to improve. According to this idea, this paper will do the following research. First through the objective of redundant association rules algorithm obtained after treatment there is no redundant association rules, and the generated association rules are classified, the users are interested in the property or rules as a goal oriented, add template idea as the user expresses the meaning of the carrier, and the classification template. The subjective interest measure is optimized to improve, so as to improve the method of calculation of interest measure.