Computational Accountability in MAS Organizations with ADOPT

• Main Authors: Matteo Baldoni, Cristina Baroglio, Katherine M. May, Roberto Micalizio, Stefano Tedeschi
• Format: Article
• Language: English
• Published: MDPI AG 2018-03-01
• Series: Applied Sciences
• Subjects: methodologies for agent-based systems; organizations and institutions; socio–technical systems; computational accountability; social commitments; agent-based programming
• Online Access: http://www.mdpi.com/2076-3417/8/4/489

This work studies how the notion of accountability can play a key role in the design and realization of distributed systems that are open and that involve autonomous agents that should harmonize their own goals with the organizational goals. The socio–technical systems that support the work inside human companies and organizations are examples of such systems. The approach that is proposed in order to pursue this purpose is set in the context of multiagent systems organizations, and relies on an explicit specification of relationships among the involved agents for capturing who is accountable to whom and for what. Such accountability relationships are created along with the agents’ operations and interactions in a shared environment. In order to guarantee accountability as a design property of the system, a specific interaction protocol is suggested. Properties of this protocol are verified, and a case study is provided consisting of an actual implementation. Finally, we discuss the impact on real-world application domains and trace possible evolutions of the proposal.