Research On Application Of Multi-Agent Coalition Formation Technology For Combinatorial Trading In Electronic Commerce

Electronic Commerce has attracted much attention for it offers significant opportunities to make businesses faster, cheaper, and more agile. However these days, it is facing severe challenge because it cannot adapt itself to users'expanding demand for personality and automation of businesses. This dissertation investigates the problem of combinatorial trading and applies multi-agent coalition formation methodology for it. From two different points of view, namely, macroscopic level and microscopic level, several key techniques of customer coalition formation in combinatorial markets are deeply studied.First, at the macroscopic level, we describe a combinatorial marketplace as a stochastic swarm system and propose a general mathematical analysis methodology for study the dynamics of this kind of multi-agent system. Based on the Rate Equation, a differential equation is derived, which gives a quantitative characteristic for coalition formation process of large-scale customers in our combinatorial market. Simulation experiments are performed to study the global behavior of the system and the effectiveness of the agent mediated combinatorial trading. Then, A new intermediary-driven combinatorial electronic market model is presented. Customer coalition formation in these marketplaces is studied.Second, from the reversed view of above research, at the microscopic level, planning ability of deliberative buyers'agents is considered. An intermediary multiagent system-driven combinatorial electronic market model IMASCM is gived. A plan representation for combinatorial trading is presented. Then, automated planning process mixed with coalition formation process is studied, which consists of demand propagation, coalition formation for single article and overpay transmit. Feasible purchasing plans are derived from the coalition planning algorithm for customers.End, based on the IMASCM model and the coalition planning algorithm, correlative theories of smart document agent and web services-based tuplecenters organization are investigated. Using them as agents'framework and communication model, architecture of the third-party platform in combinatorial markets is designed. And an experimental system is realized, which validates concepts and algorithms of above coalition planning.