Incremental algorithm for process mining based on sliding window

Most existing process mining algorithms were designed for static models and static event logs, so they could not be used in mining evolutionary processes. To deal with this problem, an incremental mining algorithm was proposed, which applied a sliding window to event log stream. And event-relation count and event-relation threshold mechanism were introduced by applying log event-relation model. The unremitting mining of event log flow was realized and a series of models corresponding to evolutionary event logs were obtained. Algorithm property and relevant parameters effect were also analyzed. Experiments were performed to validate the proposed algorithm.