Multi-issue negotiation protocol for agents: exploring nonlinear utility spaces

  • Authors:

  • Takayuki Ito;Hiromitsu Hattori;Mark Klein

  • Affiliations:

  • Department of Computer Science, Department of Techno-Business Administration, Nagoya Institute of Technology;Center for Collective Intelligence, Sloan School of Management, Massachusetts Institute of Technology;Center for Collective Intelligence, Sloan School of Management, Massachusetts Institute of Technology

  • Venue:
  • IJCAI'07 Proceedings of the 20th international joint conference on Artifical intelligence
  • Year:
  • 2007

Quantified Score

Hi-index 0.00

Visualization

Abstract

Multi-issue negotiation protocols have been studied widely and represent a promising field since most negotiation problems in the real world involve interdependent multiple issues. The vast majority of this work has assumed that negotiation issues are independent, so agents can aggregate the utilities of the issue values by simple summation, producing linear utility functions. In the real world, however, such aggregations are often unrealistic. We cannot, for example, just add up the value of car's carburetor and the value of car's engine when engineers negotiate over the design a car. These value of these choices are interdependent, resulting in nonlinear utility functions. In this paper, we address this important gap in current negotiation techniques. We propose a negotiation protocol where agents employ adjusted sampling to generate proposals, and a bidding-based mechanism is used to find social-welfare maximizing deals. Our experimental results show that our method substantially outperforms existing methods in large non-linear utility spaces like those found in real world contexts.