Mining High Utility Sequential Patterns in Dynamic Databases

Abstract

High-utility sequential pattern mining(HUSPM) has become a popular problem in the field of data mining. Many algorithms have been designed for mining high-utility sequential patterns(HUSPs), but most of the sets deal with static databases. In dynamic database mining, whenever new data comes in, the entire database needs to be rescanned to update the acquired information, thus taking up a lot of time and resources in the process of maintaining and updating the discovered information. In order to solve this problem, in this paper, we propose an incremental mining algorithm called Pre-HUSPM, based on the concept of pre-large for inserting new sequences in dynamic databases to maintain the discovered high-utility sequential patterns.