Research On Mining Algorithm Of All-Strong-Pairs On Relation Database

Abstraction mining in large transactional database is one of the most well-studied problems in data mining. However, it is well known that in the traditional association mining based on support, the association among items may not be found. Meanwhile too many rules without real association can merge. Hence, statistic association has been applied by more and more researchers. Recently, the mining of all-strong-pairs with statistic correlations on transactional database is highlighted. Given a minimum correlation threshold G and the transactional database to be mined, all-strong-pairs correlation query finds all item pairs with Pearson's correlation coefficients above 0 .Meanwhile, most data mining is about relation database since a lot of transactional data is stored in relation database. Therefore, the mining of all-strong-pairs on relation database has a large sense of research.Main topics include:1) In order to reduce the computation cost of candidate pairs, we have developed the Taper algorithm according to INF property. The developed TaperR algorithm can cut the number of candidate pairs to improve efficiency. Experimental results exhibit that the new algorithm is well-worked in the mining of all-strong-pairs. So it is more suitable for real relation database system.2) For the mining of Top-K all-strong-pairs on relation database, we propose Top-K all-strong-pairs algorithm based on threshold-estimating. Experimental results verify the effect of new method.