Event Pattern Matching with Interval Conditions

The patent segments the event pattern matching process into distinct phases: generating candidate patterns from event streams, filtering candidates based on interval conditions, and verifying final patterns. This segmentation allows the system to handle noise events at each stage rather than being overwhelmed by them in a single pass, thereby improving detection accuracy without losing valid patterns.

The patent performs preliminary filtering of event streams before full pattern matching by pre-processing events to identify potential pattern candidates that meet basic interval conditions. This preliminary action reduces the volume of noise events that proceed to full matching, improving both accuracy and efficiency while preserving all potential valid patterns.

The patent implements a dynamic pattern matching approach where the automaton structure adapts based on the specific event stream characteristics and pattern requirements. Rather than using a fixed complex automaton for all cases, the system dynamically generates or selects appropriate automaton configurations, improving efficiency for parallel and selective patterns while managing complexity through adaptive structuring.

The patent changes key parameters of the matching process including interval condition thresholds, pattern candidate windows, and automaton state transitions based on the specific requirements of parallel and selective patterns. By adjusting these parameters dynamically, the system achieves efficient matching without requiring permanently complex automaton structures for all pattern types.

The patent applies local quality by storing reception times selectively rather than uniformly for all events. Specifically, it stores reception times only for events that are potential pattern candidates or for events at critical decision points in the matching process. This localized storage approach maintains verification accuracy for interval conditions while significantly reducing the total quantity of stored reception time data.

The patent extracts and stores only the essential reception time information needed for interval condition verification, rather than storing complete event data. By extracting just the timing components necessary for pattern matching and discarding redundant information, the system achieves precise interval verification with minimal data storage requirements.