The Continuable Mining Approach to Mining Frequent Serial Episodes in the Event Sequence

摘要

In this paper, a novel approach to mining frequent serial episodes from streams of time series, is proposed. The approach is called TFSE(i.e., Time-table-joined Frequent Serial Episodes). And the conception of Episode-timetable and re-mine is put forward. An Episode-time-table is corresponding with a episode pattern. The Episode-timetable records the start-time and end-time of all of the corresponding episodes. A new Episode-time-table is generated by joining two appropriate Episode-time-tables. The amount of records in the Episode-time-table is the number of episodes occurring in the time series. Re-mine need not to scan original time series and retrench much resource and time. Because the middle results are stored and the measure of joining two appropriate Episode-time-tables is taken to test candidate episode, to a certain extent, TFSE reduces difficulty with mining long episodes.