METHOD FOR PARALLEL MINING OF TEMPORAL RELATIONS IN LARGE EVENT FILE

摘要

A method for parallel exploration of temporal relationships in large event files using a MapReduce model is disclosed. The method for parallel exploration of a time relationship in a large event file according to the present invention comprises the steps of: arranging an event file based on a customer ID and an event time when an event occurs, and a type of frequent events satisfying more than a predetermined support degree in the event file. Generating a set, converting the event file into a frequent event sequence including the frequent event type set, generating a time interval file by summarizing the frequent event sequence, and generating a candidate time relationship in the time interval file. And generating a frequent time relationship that satisfies a predetermined support degree or more in the candidate time relationship, and generating a time relationship rule from the frequent time relationship.