Extending the RETE algorithm for event management

A growing number of industrial applications use rule-based programming. Frequently, the implementation of the inference engine embedded in these applications is based on the RETE algorithm. Some applications supervise a flow of events in which time, through the occurrence dates of the events, plays an important role. These applications need to be able to recognize patterns involving events. However the RETE algorithm does not provide support for the expression of time-sensitive patterns. This paper proposes an extension of RETE through the concepts of time-stamped events and temporal constraints between events. This allows applications to write rules that process both facts and events.