Top-K high utility episode mining in complex event sequence

Abstract

Mining high utility episodes in complex event sequences is an emerging topic in data mining. In utility mining, users set a minimum threshold and the episodes having higher utility than the threshold are reported. The utility framework introduced in episode mining provides more informative and usable knowledge as compared to frequent episode mining. However, it is difficult for the user to set an appropriate minimum utility thresh- old. As the user cannot predict the count of mined episodes by the utility threshold, the number of reported episodes can vary hugely in accordance to the set threshold. To address this issue, in this thesis we propose an algorithm for mining Top-K high utility episode in complex event sequence. It discovers episodes with highest utility to ones with kth highest utility where the user can set the desired count of episodes, k. We also propose two different strategies to reduce the search space by raising the minimum threshold e ectively. We conduct experiments on real dataset and show the effectiveness of our approach.