The event calculus at work: a case study in the medical domain

Efficiency issues involved in temporal reasoning have often been ignored, and this is a major limiting factor to the application of temporal reasoning in realistic domains. In this paper we adopt the event calculus (EC) model of time and change, enriched with preconditions. The addition of preconditions, which is crucial to deal with real-world examples, heavily deteriorates the performance. Therefore, we extend EC with a caching mechanism, called cached event calculus (CEC), that improves efficiency and preserves the requirements of EC; in particular, it makes no assumptions about the temporal order of input events. We first introduce EC, and extend it with preconditions, using the PROLOG language. Then, we present CEC, and compare the complexity of query and update processing in EC and CEC. We show how CEC scales up to real-world problems by illustrating its application to a patient monitoring task. Finally, we analyse the experimental results, provide an assessment of the work, and outline future developments