Formalizing and achieving multiparty agreements via commitments

  • Authors:
  • Feng Wan;Munindar P. Singh

  • Affiliations:
  • North Carolina State University, Raleigh, NC;North Carolina State University, Raleigh, NC

  • Venue:
  • Proceedings of the fourth international joint conference on Autonomous agents and multiagent systems
  • Year:
  • 2005

Quantified Score

Hi-index 0.00

Visualization

Abstract

Multiparty agreements often arise in a multiagent system where autonomous agents interact with each other to achieve a global goal. Multiparty agreements are traditionally represented by messaging protocols or event-condition-action rule sets in which agents exchange messages in a predefined sequence to ensure both global and local consistency. However, these models do not readily incorporate agents' autonomy and heterogeneity, which limits their ability to help build a flexible open system. Commitments have been studied for modelling various agent interactions. They have also been used as the key elements for formulating multi-party agreements and centralized approaches for resolving potential conflicts. This paper extends the above results by refining the formalizations and the existing protocols and proposing a decentralized protocol which is more efficient in resolving conflicts. It also introduces the concept of protocol safety, which ensures that agents not only interact efficiently but also correctly. This approach is geared toward constructing business processes where agents are mutually constraints in a manner that preserves their autonomy and heterogeneity.