In agent-based applications in electronic commerce and other information systems, agents, representing real-life negotiating parties, negotiate resources and services with other agents or humans. Self-interested software agents, often with conflicting preferences, can negotiate mutually acceptable agreements. In this dissertation, we focus on the following major aspects of such automated negotiations: negotiation frameworks, negotiation strategies, effects of social context and agent attitudes on negotiation outcomes. Additionally, we discuss an agent architecture for effectively using arguments for negotiation and study an automated supply chain as an example application of automated contracting. The primary contributions of this dissertation are the following: (i) the development of efficient frameworks for multi-issue negotiation with divisible and indivisible issues, (ii) the design of a negotiation strategy that models opponent's preferences in repeated negotiations to improve individual utility of the agent, (iii) the analysis of the performance of multiple negotiation strategies under different negotiation environments and their evolution in a society of adaptive agents, and, (iv) the development of an architecture and a decision theoretic framework for argumentation-based negotiations. We have also designed an efficient scheduling and pricing strategy for a supplier in a dynamic electronic supply chain, an application of automated contracting, where contracts are awarded via an auction-based negotiation protocol.
展开▼
