Multi-agent system for consumer-oriented electronic commerce

Abstract

With the advent of the information superhighway and the exponential growth of the Internet usage, the importance of multi-agent systems is proliferating. The central theme of this thesis is to demonstrate the benefits of adopting multi-agent system (MAS) paradigm to implement consumer oriented electronic commerce system. The discipline of computational science is exploited to provide insights into the behaviour of a model of consumer behaviour that reflect the cognitive notion that the thesis has developed. For this, a multi-agent system computational environment is used to model and investigate the consumer purchase over the Internet. The MAS is developed based on a presented taxonomy, that is most relevant to the thesis application. The thesis also presents a novel approach to negotiation. Results of empirical evaluations provide a strong support that agents using the proposed approach would achieve higher payoff than human subjects. An empirical evaluation for the usability of the prototype system is also presented. Reported results are very encouraging to implement a fieldable system. To complement the perspective for a complete consumer-oriented EC system, the thesis addresses and develops approaches for searching and extracting relevant information. Example experiments are also reported to act as indicators for the effectiveness of the developed approaches.