Discovering injective episodes with general partial orders

摘要: Frequent episode discovery is a popular framework for temporal pattern in event streams. An partially ordered set of nodes with each node associated an type. Currently algorithms exist only when the partial order total (serial episode) or trivial (parallel episode). In this paper, we propose efficient discovering frequent episodes unrestricted orders event-types are unique. These can be easily specialized to discover serial parallel episodes. Also, flexible enough mining space certain interesting subclasses orders. We point out that frequency alone not sufficient measure interestingness context mining. new which, used along frequency, results scheme data Simulations presented demonstrate effectiveness our algorithms.