Two Approaches to Event Detection in Active Database Systems

An active database management system can react to predefined events in the database, resulting in performing some appropriate action. ECA-rules are used to capture these predefined events and to express the appropriate action in the active database management system. An event is said to be an atomic occurrence and is either primitive or composite. D-SNOOP and O-SNOOP are two ways to describe the semantics of composite events. D-SNOOP is detection-based (an event is considered to be instantaneous) and O-SNOOP occurrence-based (an event is considered to occur over a time interval). Some problems concerning the detection of composite events using D-SNOOP semantics have been highlighted by Galton and Augusto (2001), who have created the O-SNOOP semantics to rectify some of the shortcomings of D-SNOOP. It is, however, not known what practical consequences O-SNOOP has on applications. To find some of these practical consequences, an advanced application that uses composite events must be used. In this work, the advanced application is a cooperative information system. A cooperative information system is a system in which several agents work together to solve some common problem. The agents can take on a role as either manager (service requester) or agent (problem solver). In this dissertation an investigation of the differences between O-SNOOP and D-SNOOP in applications handling cooperation protocols is presented. Our first objective has been to model a cooperation protocol using in turn the D-SNOOP and the O-SNOOP semantics. A further objective has been to compare the models of the cooperation protocols to identify any differences between using D-SNOOP and O-SNOOP semantics in applications handling cooperation protocols