Link to publisher version (URL)
Within an agent society, agents utilise their knowledge differently to achieve their individual or joint goals. Agent negotiation provides an effective solution to help agents reach agreements on their future behaviour in the society to guarantee their goals can be achieved successfully. Agents may need to conduct Multiple Interdependent Negotiations (MIN) with different opponents and for different purposes, in order to achieve a goal. By considering the complexity of negotiation environments, interdependencies, opponents and issues in the agent society, conducting MIX efficiently Is a challenging research issue. To the best of the authors' knowledge, most of the state-of-art work primarily focuses on single negotiation scenarios and tries to propose sophisticated negotiation protocols and strategies to help individual agents to succeed in single negotiation. However, very little work has been done while considering interdependencies and tradeoffs among multiple negotiations, so as to help both individual agents as well as the agent society, to increase their welfare. This paper promotes the research on agent negotiation from the single negotiation level to the multiple negotiations level. To effectively conduct MIN in an agent society, this paper proposes three feasible negotiation procedures, which attempt to conduct MIN in a successive way, in a concurrent way, and in a clustered way while considering them in different negotiation situations, respectively. A simulated agent society is built to test the proposed negotiation procedures with rand om experimental settings. According to the experimental results, the successive negotiation procedure produces the highest time efficiency, the concurrent negotiation procedure promises the highest profits and success rates, whilst the clustered negotiation procedure provides a well-balanced solution between negotiation efficiency and effectiveness.