Research Of Positive And Negative Association Rules Mining Technology Based On Multiple Supports

Association rule (AR) is one of the most important techniques of data mining, the existing association rules mining algorithms are most based on frequent itemsets, the researchs about infrequent itemsets are very little. However when we study the negative association rules, infrequent itemsets become more and more important because they contain a lot important negative association rules. At the same time, the existing algorithms of mining association rules are almost achieved by the single minimum support, but in fact the occurrence frequencies of the events corresponding to each attribute are very different, so the restriction of single minimum support is unable to reflect the objective.For the sake of offseting the shortage of single minimum support, multiple minimum support algorithms appear. This article brings forward three new algorithms based on multiple minimum support algorithms which ameliorate and perfect the old multiple minimum support algorithms. The three new algorithms are MMS-inFS algorithm, 2LMS-inFS-FS algorithm and MLMS algorithm.The MMS-inFS algorithm adds a restriction of infrequent itemsets based on multiple supports algorithm which makes it can mines both infrequent itemsets and frequent itemsets simultaneous. The 2LMS-inFS-FS algorithm is based on two level multiple supports, it ameliorates the MMS-inFS algorithm which makes a restriction of infrequent itemsets from single support to multiple supports. It sets two level supports for each item, one is used to restrict infrequent itemsets and the other is used to restrict frequent itemsets. In this way, it can restrict frequent itemsets and infrequent itemsets better. The MLMS algorithm is based on multiple layer minimum supports which sets different minimum supports for different items.For the new algorithms, we compared them with several other algorithm models and we do some experimentions to prove the validity of these new algorithms.