Article Preview
Top1. Introduction
Normative multi-agent systems as the result of combining two established fields: normative systems and multi-agent systems (Mahmoud et al., 2014) are an extremely active research area (Morales et al.,2014). Normative systems are ones that their behaviours are described and specified using norms (Boella et al., 2006). By norms, we mean “rules and standards that are understood by members of a group and that guide and/or constraint social behaviours” (Cialdini & Trost, 1998). Knowing that multi-agent systems are composed of a set of interacting entities (called agents), normative multi-agent systems can be defined as multi-agent systems such that the behaviours of the agents and the interaction between them are specified by means of norms (Cardoso, 2007; Viganò & Colombetti, 2009). These norms can be violated by agents to deviate from the ideal behaviour instead of classical computer programs that model only the ideal behaviour of systems (Broersen et al., 2001; Boella & van der Torre, 2004).
Describing multi-agent systems using norm concepts provides several advantages like simplifying the cooperation and coordination among a group of agents, enhancing the flexibility and the autonomy of agents and promoting openness and heterogeneity (Mashayekhi et al., 2016; Huang et al., 2016). Nevertheless, studying normative multi-agent systems still faces several challenges that require further investigation. In fact, implementing normative multi-agent systems is one of these challenges. Indeed, many recent works have proposed solutions to implement normative multi-agent systems (Grossi et al., 2010; Morales et al., 2014; Hübner et al., 2011; Dastani et al., 2009; Kafali et al., 2017). However, most of these proposals suffer from many deficiencies. As examples of these deficiencies, we can mention the need for deep knowledge of some formalisms, the need for deep knowledge of some organisational models or the difficulty of expressing the whole concepts of normative systems. Several challenges of implementing normative multi-agent systems need more investigation like norms creation, norms representation, norms sharing and transmitting, norms updating, reasoning about norms, the monitoring mechanism and the regulation mechanism (Mahmoud et al., 2014; Grossi et al., 2010.
In this work, a new open source framework called NorJADE (NorJADE, 2019) is proposed to support the implementation of normative multi-agent systems. This framework is based on the most popular open source JADE platform (Bellifemine et al., 2005). The extensions are made using two established software technologies namely, the ontology and aspect-oriented programming (Kiczales et al., 1997). Thanks to the used technologies, the proposed framework provides several advantages. The ontology allows for an exogenously, explicit and sharable representation of norms. Moreover, it simplifies the automatic modification of existing norms dynamically (i.e. at the runtime). On the other hand, we use aspect-oriented programming to extend the established JADE platform with the required mechanisms allowing for the manipulation and reasoning about norms. Using already exist and popular multi-agent platform to develop normative multi-agent systems can simplify the developers task. Finally, the developed framework is an open source project and is available on (NorJADE, 2019).