Sigon: A multi-context system framework for intelligent agents. (1st April 2019)

• Record Type: Journal Article
• Authors: Gelaim, Thiago Ângelo; Hofer, Valdir Luiz; Marchi, Jerusa; Silveira, Ricardo Azambuja
• Is Part Of: Expert systems with applications. Volume 119(2019)
• Page Start: 51
• Page End: 60
• Publication Date: 2019-04-01
• Subjects: Agent -- Multi-context systems -- Belief-Desire-Intention; Expert systems (Computer science) -- Periodicals; Systèmes experts (Informatique) -- Périodiques; Electronic journals; 006.33
• DOI: 10.1016/j.eswa.2018.10.042
• Languages: English
• ISSNs: 0957-4174

Abstract

Highlights: A framework for the development of agents as multi-context systems. Programming BDI agents as multi-context systems and extending it with new features. The agent development flexibility and performance through bridge rules. Abstract: The usage of agents has become one of the most effective ways to deal with complex systems. The agent's approach allows for the parallel execution of actions and eases the conception of the system. In this paper, we introduce a language, called Sigon, that allows the definition of agents as multi-context systems. Multi-context systems, in turn, allow for a modular description of the agent, and the implementation of bridge rules contributes to the system flexibility. We propose a generic framework to conceive agents, in which the system's developer defines their internals in terms of contexts and bridge rules. In order to show the way in which the programmer can aggregate other logical contexts, we first present the configurations of a BDI-like agent in our framework. Subsequently, the programmer can add more contexts, such as emotional or negotiating ones, and develop other agent architectures. Thus, our approach can be characterized as a form of generic and extensible architecture. In order to validate our proposal, we introduce some examples that delineate some strengths of the proposal, and also we present an analysis of bridge rules execution to demonstrate the processing overhead.