A robust skip-till-next-match selection strategy for event pattern matching

Abstract

In event pattern matching, various selection strategies have been proposed to impose additional constraints on the events that participate in a match. The skip-till-next-match selection strategy is used in scenarios where some incoming events are noise and therefore should be ignored. Skip-till-next-match is prone to blocking noise, i.e., noise that prevents the detection of matches. In this paper, we propose the robust skip-till-next-match selection strategy, which is robust against noise and finds matches that are missed by skip-till-next-match when blocking noise occurs in the input stream. To implement the new strategy in automaton-based pattern matching algorithms, we propose a backtracking mechanism. Extensive experiments using real-world data and different event pattern matching algorithms show that with skip-till-next-match the number of matches not detected due to blocking noise can be substantial, and that our backtracking mechanism outperforms alternative solutions that first produce a superset of the result followed by a post processing step to filter out non-compliant matches.