When to Update the sequential patterns of stream data?

TL;DR

This paper introduces a method to determine optimal update timing for sequential patterns in streaming data by balancing performance and pattern difference, validated through experiments with the IUS algorithm.

Contribution

It defines a new difference measure for sequential patterns and proposes the TPD method to decide update timing based on performance-difference tradeoff.

Findings

Speedup decreases as window size increases.

Difference measure increases with window size.

Incremental ratio varies between 20% and 30% for IUS.

Abstract

In this paper, we first define a difference measure between the old and new sequential patterns of stream data, which is proved to be a distance. Then we propose an experimental method, called TPD (Tradeoff between Performance and Difference), to decide when to update the sequential patterns of stream data by making a tradeoff between the performance of increasingly updating algorithms and the difference of sequential patterns. The experiments for the incremental updating algorithm IUS on two data sets show that generally, as the size of incremental windows grows, the values of the speedup and the values of the difference will decrease and increase respectively. It is also shown experimentally that the incremental ratio determined by the TPD method does not monotonically increase or decrease but changes in a range between 20 and 30 percentage for the IUS algorithm.