On progressive sequential pattern mining

Abstract

When sequential patterns are generated, the newly arriving patterns may not be identified as frequent sequential patterns due to the existence of old data and sequences. In practice, users are usually more interested in the recent data than the old ones. To capture the dynamic nature of data addition and deletion, we propose a general model of sequential pattern mining with a progressive database. In addition, we present a progressive concept to progressively discover sequential patterns in recent time period of interest.

A taxonomy of sequential pattern mining algorithms

“…(1) Sampling and/or compression: Compression is mainly used in pattern-growth tree projection methods, such as FS-Miner [El-Sayed et al 2004]; WAP-tree ; PLWAP [Lu and Ezeife 2003]; and PS-tree (a progressive sequential-mining algorithm [Huang et al 2006]). In the apriori-based category, compression is used to solve the problem of fast memory consumption due to the explosive growth of candidate sequences by some apriori-based algorithms, like Apriori-GST [Tanasa 2005].…”

Mining sequential patterns across multiple sequence databases

“…To prevent candidate generation, DISC-all [8] used a novel sequence comparison strategy. A progressive concept has been explored in mining sequential patterns to capture the dynamic nature of data addition and deletion [15]. The above research works are focused on improving the performance of traditional sequential pattern mining.…”