Mining Recent High-Utility Patterns from Temporal Databases with Time-Sensitive Constraint

摘要

Useful knowledge embedded in a database is likely to be changed over time. Identifying recent changes and up-to-date information in temporal databases can provide valuable information. In this paper, we address this issue by introducing a novel framework, named recent high-utility pattern mining from temporal databases with time-sensitive constraint (RHUPM) to mine the desired patterns based on user-specified minimum recency and minimum utility thresholds. An efficient tree-based algorithm called RUP, the global and conditional downward closure (GDC and CDC) properties in the recency-utility (RU)-tree are proposed. Moreover, the vertical compact recency-utility (RU)-list structure is adopted to store necessary information for later mining process. The developed RUP algorithm can recursively discover recent HUPs; the computational cost and memory usage can be greatly reduced without candidate generation. Several pruning strategies are also designed to speed up the computation and reduce the search space for mining the required information.