![]() |
![]() |
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
![]() ![]() ![]() ![]() ![]() |
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
![]() |
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Home | Research | Research Programs | Internet Logic |
Internet LogicMarketPlace NB: Increased utility of multi-agent e-markets through enriched protocols and adaptive decision proceduresThe project started in 2002 and is ongoing. Starting from the simple e-shopping protocol, where a consumer of a product or service asks a supplier for a price quote, we suggest an additional optional message in the protocol: We allow the supplier to give the consumer a time frame when a price will be valid, and a probability distribution function from which the price, a random variable, will be assigned. This offers a benefit to the supplier; he does not need to set a price until after he has a chance to observe market demand. We have also measured an increase in the utility that a consumer can expect, given that the consumer's decision of which offer to accept takes these price estimates into account. We aim to further enhance protocols and to assess for each business partner the expected benefits, defined as increased expected utility of the transaction. Note that protocols apply both to inter-agent and multi-agent systems. Expected utility is measured two ways: theoretically based on statistical models, and experimentally based on simulations. For the simulations we already have a prototype, that places given numbers of buyers and of sellers into an e-market with randomly selected utility functions for the goods they possess, and the goods they desire. Buyers and sellers interactions as defined by the protocol we are evaluating. With each addition to the protocol, we need to enrich the decision procedures in the agents, to take advantage of the new information. We observe how the market increases in total utility for all agents, as a function of the number of interactions. Extensions to these experiments are planned where re-selling and coalition-managing agents are introduced. Our research results could have an impact on the following sectors: Electronic commerce, the electronic marketplace, multi-agent systems for e-commerce. Research opportunities related to this project: Multi-agent systems, auction servers Possible applications include: e-marketplaces, electronic negotiations Research ContactDr. Bruce Spencer Business ContactMarc-Alain Mallet |
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
![]() |
|