Mining frequent items in a product of partial orders using parallel calculations

We consider the issues of data analysis with items from the Cartesian product of finite partially ordered sets. For efficiently mining frequent items generated by all possible binarization options for the original nonbinary data, we use a modification of the classic FP-tree (Frequent Pattern Tree). Time savings are achieved through the use of parallel computing based on Compute Unified Device Architecture (CUDA) technology. The results of testing the constructed parallel procedures for synthesizing the desired frequent items using model and real data are presented.