| Recently, because Internet auctions have garnered a great deal of interest, it has become an especially popular trading mode of Electronic Commerce (EC) and a promising field for applying agent and Artificial Intelligence Technologies. Various theoretical and practical studies on Internet auctions have already been conducted. Among these studies, those on multidimensional-auctions have lately attracted considerable attention. However, multidimensional-auctions have also lack of integrity systematic theoretical support and practical guidance. On the one hand, as a result of the complexity of multi-dimensional auction itself, theoretical research has lagged behind, operating efficiency and the complexity of practice have been the bottleneck problem to its extensive usage; On the other hand, despite the rapid development of information technology makes multiple attributes online auctions become possible, there are still many problems and difficulties in operation. Optimize multi-dimensional online auction mechanism and decision-making model are conducted in this context that through the use of intelligent computing technology to set up more complex auction trading rules for the organizations and programs to achieve more complex auctions.The selection of winning bids in the first-prices multidimensional-auctions is an extremely complex problem; in fact it has been shown to be NP-complete. Our research problem is to introduce meta-heuristic algorithms to solve the identical problem. The meta-heuristic algorithms put the problem to be solved into a proper science perspective. The primary motivation for studying this problem was an acknowledgement of the need for combinatorial auction mechanism that more accurately reflects the demands of the bidders. Secondly, the ability to harness computational intelligence has made possible more complex combinatorial reverse auction mechanism opening an exciting area of research and the models available has been expanded through this endeavor. Finally, the implementation of the models and algorithms in the combinatorial tendering and bidding software system provides a way to put theory into practice.The major progresses of this paper are shown as follows:Firstly, this dissertation summarizes and analyzes the contents and methods of traditional auction theory, which leads to online auctions this scientific problem. To reference point model for the foundation, through the analysis of basic principles to optimal online auction mechanism design, pointing out the inadequacy of the original method, then provide strategy learning method based on intelligent agent. Based on the analysis to the basic theory of online auctions, the auction space will be extended to multi-dimensional space. The first question is multi-attribute auction to unit item. this dissertation analyze the nature and build a multi-attribute to single-item model for online auctions, and conduct a detailed analysis to bidders strategy, cost function, and the optimal score function; then relax the assumption and take more analysis to multi-attribute to unit item online auctions.Having fully researched on multi-attributes online auction to unit item, the auction space will be expanded, namely discuss the multi-attributes online auction model to mluti-items, and classify the items to homogeneous and heterogeneous according to the quality, and obtain the generally conclusion. In homogeneous multi-dimensional model, what needs to do is the analysis to optimal mechanism and buy; in the heterogeneous multi-dimensional model, what needs to do is the description to main the incentive compatibility and individual factors theories, and to the characteristics of online auctions. It can provide dynamic optimization mechanism for multi-dimensional online auction, and analyze the score function of dynamic optimization mechanism, bidding rules, as well as the bidding strategy.Because of the fuzzy character of decision relation, this dissertation presents a fuzzy rough Study on optimization algorithms of winner determination in multi-dimensional auctions. Beginning with combinatorial auctions mechanisms in the electronic commerce, the complexity of the set of target and winner determination is analyzed on the background of the first-rice sealed auction. The general model of winner determination in multi-dimensional auctions is formulated. The model indicated that it is a problem on combinatorial optimization. Adopting self-crossing genetic operators and embedding the Fitting-First heuristic rules presented a genetic algorithm presented on the basis of the ideas of heuristic logarithms. Meanwhile, utilizing the characteristic of attractors in chaos, the fitting-first heuristics embedded chaotic search is also proposed.Lastly, based on the work outlined above, a prototype system of e-commerce oriented multi-dimensional is analyzed, designed and achieved. The prototype system verifies the conclusion above. This dissertation enriches and fulfills the theory of multi-dimensional online auction, overcomes the shortcomings of existed methods. The further research direction of multi-dimensional online auction is showed clearly in this dissertation. Its results will contribute greatly to future applications.
|
PDF Full Text Request