Design Of Association Rules Mining Algorithm Based On The Discriminative Concept Lattice

Association rules is one of the most knowledge representation methods, frequent pattern mining as a fundamental data mining task has widespread applications in many different domains. Concept lattice is a very useful formal analysis tool and can show the relationship among the concepts vividly and briefly. In addition, every node in the concept lattices is a frequent itemsets, and there is a one-to-one correspondence between concept intensions and frequent itemsets. Abundant literate has been conducted in-depth research in mining frequent itemsets and association rules based on the concept lattice. However, most of them did not take into account the differences of attributes when concept lattice is build. There are two obvious deficiencies based on this idea: (1) It will lead to the issues of combinatorial explosion and redundancy; (2) It need a long time to construct the concept lattice and low efficiency, because we did not take into account the differences of attributes in the process of building concept lattice.In this paper, i propose a new concept called discriminative intension. Every attributes has a discriminative power (DISP). In the process of building concept lattice, by removing the attributes of low discriminative power, It will reduce the number of frequent itemsets, then speed up the step of constructing the lattice; Next, there has been a new method to calculate the DISP and reset the value under some condition, and scan the database on each layer, by which it can reduce the times of scanning the database, then decrease the time of generate association rules; Furthermore, by improving the algorithm of mining association rules based on the concept lattice, we introduce the concept of confidence to the mining association rules, decraese the time of calculate confidece and improve the efficiency; Finally, giving the algorithm of mining association rules based on the concept lattice of discriminative power Data minging of association rules based on the discriminative intension, it has two steps: (1) the building of concept lattice; (2) mining frequent itemsets and association rules based on the concept lattice.The main contributions are as follows:1)proposing a new concept called discriminative intension., it will reducetion the number of frequent itemsets based on this concept, then speed up the step of constructing the lattice;2)proposing a new method to calculate the DISP and reset the value under some condition, scan the database on each layer, can reduce the number of scanning the database, then decrease the time of generate association rules;3)improving the algorithm of mining association rules based on the concept lattice, introducing the concept of confidence to the mining association rules, and decraeseing the time of calculate confidece and improve the efficiency;4)giving the algorithm of mining association rules based on the concept lattice of discriminative power.