Online Constrained Pattern Detection over Streams

Online pattern detection poses a challenge in many data-intensive applications, including network traffic management, trend analysis, intrusion detection, and various intelligent sensor networks. These applications have to be time and space efficient while providing high quality answers. Meanwhile, far less attention has been paid for detecting constrained patterns, that cannot be simply matched because there is no available pattern for prediction. This paper presents our research effort in efficient pattern detection with constraint. We propose a new method named Online Pattern Detection with Constraint (OPDC) to detect constrained patterns over evolving data stream, taking into account various user-defined constraints. To ensure that the constrained patterns are representative, we extend regular expression in a simple but powerful way. Our experimental results on real data sets demonstrate the feasibility and effectiveness of the proposed scheme.