Maintenance of Prelarge High Average-Utility Patterns in Incremental Databases.

Most existing data mining algorithms focused on mining the information from the static database. In this paper, the principle of pre-large is used to update the newly discovered HAUIs and reduce the time of the rescanning process. To further improve the performance of the suggested algorithm, two new upper-bounds are also proposed to decrease the number of candidates for HAUIs. The experimental results show the proposed algorithm has excellent performance and good potential to be applied in real applications.