Mining Frequent and Top-K High Utility Time Interval-based Event with Duration Patterns from Temporal Database

摘要

Time interval-based pattern mining is proposed to improve the lack of the information of time intervals by sequential pattern mining Previous works of time interval-based pattern mining focused on the relations between events without considering the duration of each event However the same event with different time durations will cause definitely different results For example if some patients cough for one day they may get a cold for one week In contrast if some patients cough for one year they may get pneumonia in the future In this work we propose two algorithms SARA and SARS to extract the frequent Time Interval-based Event with Duration TIED patterns TIED patterns not only keep the relations between two events but also reveal the time periods when each event happens and ends In the experiments we propose a naive algorithm and modify a previous algorithm to compare the performances with SARA and SARS The experimental results show that SARA and SARS are more efficient in execution time and memory usage than other two algorithms Otherwise TIED pattern still has limitation which do not consider the utility value in each time stamp We propose another pattern High Utility Time Interval-based with Duration Pattern which combine the concept of utility pattern mining and TIED pattern mining We design an algorithm LMSpan to mine High Utility Time Interval-based with Duration HU-TIED Pattern In the experiments we design another naive algorithm Naive UDP to compare with LMSpan LMSpan take less time and memory use and generate less candidate than Naive UDP

獎項日期	2014 8月 26
原文	English
監督員	Jen-Wei Huang (Supervisor)