Parallelization of Sequential Pattern Sampling

In the last years, the field of data mining has undergone extensive work on patterns discovery by sampling techniques. Recently, these sampling methods have been applied to sequential data that are complex in nature. The complexity of these data lies in their structure, which has a notable impact on the speed of the computation which is time consuming with huge database. In this paper, we show how to use the BSP (Bulk Synchronous Parallel) programming model to improve the efficiency of sequential pattern sampling methods. Indeed, we propose a parallel algorithm that operates on sequential databases that are knowingly distributed in order to accelerate the computation time. The analyses show the positive impact of the framework on the execution time of the method.