Agent reactive capabilities in dynamic environments

Abstract

Currently, there are many intelligent agent models that provide a number of components or abstractions to solve different types of problems. However, few of them provide abstractions or concepts that allow considering how to deal with complex dynamic environments (an environment that changes continuously in terms of available resources, global behavioral rules, etc.). In this work, we propose two abstractions that provide the developer with a new way of modeling reactive agent capabilities in dynamic environments. The first abstraction focuses on how to process the environmental stimuli as events; the second abstraction specifies how to launch tasks in response to events, which is an approach that is based on event–condition–action rules. Moreover, we present an example that is based on a call center CBR-based application, and a performance evaluation of the proposal is also provided.