Handling uncertain rules in composite event systems

Abstract

In recent years, there has been an increased need for active systems - systems that are required to act automatically based on events, or changes in the environment. In many cases, the events to which the system should respond to, have to be inferred from other events based on complex temporal predicates. However, none of the existing composite event systems created to enable such inference can deal with cases in which an event cannot be inferred with absolute certainty based on the reported events. Therefore, in this-paper, we describe how a deterministic event composition system can be extended to manage such uncertainty, and specify the principles of a formal framework for such inference. The contribution of this framework is twofold: It extends the semantics of event composition in a natural manner for probabilistic settings, and it enables the application of these extensions to the quantification of the occurrence probability of events. Copyright © 2002, American Association for Artificial Intelligence (www.aaai.org). All rights reserved.