Publication: Benefits awareness-based norm adoption framework for multi-agents system
Date
2021-02
Authors
Al-Mutazbellah Khamees Itaiwi, Dr.
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
In open normative multi-agent communities, agents adopt new norms to increase their
utilities. Several studies have developed mechanisms for agents to adopt new norms,
which are based on sanction, imitation, or social learning. The limitation of these
mechanisms is that over time, all agents adopt new norms, while in real-life situations,
there are always a number of agents that persistently reject the norms for their benefits.
This work advocates that intelligent agents should adopt new norms based on their
awareness of expected benefits of the norms on their utilities, and not just by sanctions
or imitating other agents. This thesis presents the concept of norms benefit awareness
which is inspired by, and extends the concept of norms detection in open normative
multi-agent communities. In such communities, agents adopt the norms that are not
explicitly given and agents have free and unrestricted movement from one community
to another to achieve their goal. In such situations, agents must be equipped with
detection tools to adapt to the open communities. An intelligent agent should be able
to autonomously evaluate its ‘awareness’ of the norms benefits and recognize the
positive norms from the negative ones, while acting in situations it has never
experienced before. Failure to detect or discover a society’s norms could subject the
agent to probable sanction or penalty if it rejects the regulative norms. On the other
hand, succeeding in discovering the norms earns the agent rewards, or at least avoids
penalties. If a visitor agent is equipped with mechanisms to detect norms of a system,
then it becomes more flexible and is able to adapt itself to changes in the environment
or its goals. Consequently, this thesis presents a conceptual framework for agents’
‘awareness’ of norms benefits in open normative multi-agent communities. The work
on norms benefit awareness is a relatively new concept in terms of the approach used
in multi-agent systems. The thesis shows an approach of interpreting and modulating
the agent’s behaviour. In the proposed framework, four parameters constitute agents’
awareness of the norms benefits, namely Norm’s Yields; Norm’s Morality; Norm’s
Trust, and Norm’s Adoption Ratio. Using these parameters agents are able to evaluate
the benefits of observed norms and subsequently determine whether the norms
increase or decrease their utilities for eventual adoption or rejection. The norms benefit
awareness is simulated to validate the applicability and functionality of the awareness
of norms benefit framework. Simulation results show a favourable application of the
techniques and theory in open multi-agent communities.