| 摘要 |
PROBLEM TO BE SOLVED: To efficiently extract correlation from large-scale time sequential data by extracting the start time of a time sequential part as the attribute information of an event and extracting the correlation of generation tendency between the events based on the attribute information of the event. SOLUTION: Plural featured time sequential part extraction parts 12 extract the start time of the featured time sequential part from respective time sequential data bases 11, replace it with a prescribed event, convert the time sequential data into event strings and preserve the event strings in respective event string storage parts 13. Then, for the event strings, the combination of the events is extracted in an event combination extraction part 14, the frequency distribution of the start time is checked in an event duration judgement part 15, and when prescribed conditions are satisfied, extraction is performed as the candidate of the events with the correlation. Thus, the correlation is efficiently extracted from the large-scale time sequential data. |
