Efficiently Mining Gapped and Window Constraint Frequent Sequential Patterns

Sequential pattern mining is one of the most widespread data mining tasks with several real-life decision-making applications. In this mining process, constraints were added to improve the mining efficiency for discovering patterns meeting specific user requirements. Therefore, the temporal constraints, in particular, those that arise from the implicit temporality of sequential patterns, will have the ability to efficiently apply temporary restrictions such as, window and gap constraints. In this paper, we propose a novel window and gap constrained algorithms based on the well-known PrefixSpan algorithm. For this purpose, we introduce the virtual multiplication operation aiming for a generalized window mining algorithm that preserves other constraints. We also extend the PrefixSpan Pseudo-Projection algorithm to mining patterns under the gap-constraint. Our performance study shows that these extensions have the same time complexity as PrefixSpan and good linear scalability.