Tokuro Matsuo

An optimal coalition formation among buyer agents based on a genetic algorithm

Group buying is a form of electronic commerce that is growing quickly. There are many group buying sites on the Internet. Group buying is a commercial transaction in which the unit price of goods changes with the number of buyers, and a buyer can purchase goods at a low price if many buyers participate in group buying. There are several group-buying sites that are selling similar (or the same) goods. Buyers are often distributed among these group-buying sites non-optimally. If we can optimally allocate buyers to several group buying sites, all buyers can buy a good at a lower price. possible that a participant can purchase the target or the sim The aim of this paper is to solve this optimal allocation problem by a Genetic Algorithm. Our method can effectively avoid the growth of fatal genes. In experiments, we compared our method with an exhaustive search algorithm and a brute search algorithm. The experimental results show that our algorithm can optimally allocate buyers in an efficient time.

A volume discount-based allocation mechanism in group buying

Volume discount is seen as an effective form of electronic commerce and a promising field for applying agent technologies. In current volume discount mechanisms, items are not allocated efficiently to buyers. Namely, social surplus is not maximum in existing volume discount schemes. To solve this problem, we propose a volume discount mechanism based on the sellers reservation price and the payment adjustment value. First, a seller registers his/her items with the evaluation value functions. The sellers evaluation value is sealed and each buyer bids his/her evaluation value as sealed bid. After the deadline, the mechanism determines the allocation of bundles of items. A tentative price is decided and the payment adjustment value is calculated. Finally, the payment amount is calculated. Our mechanism has some key advantages. First, the mechanism is Pareto efficient. Second, our mechanism is a strategy-proof mechanism, that is, it has the incentive compatibility. Third, our mechanism provides individual rationality. Fourth, our mechanism is made based on the volume discount system, where the seller can give a signal indicating a discount for buyers. Finally, our mechanism weakens the influence of false name bids.

A Web-based Cooperative Research Paper Edit System

We propose an interactive edit Web system, that is, users can edit and revise documents cooperatively and interactively. The Ie-Web is constructed based on conceptions of special issue edit and pantographic issue edit. The goal of this study is not only developing cooperative document making system by many users, but research paper making system by mentors and other students. In recent years, computer-based tele-learning systems are developed. Thus, we propose a useful cooperative document making support system to develop a system used at university educations

An E-learning support system based on qualitative simulations for assisting consumers' decision making

One of the difficulties of using Artificial Neural Networks (ANNs) to estimate atmospheric temperature is the large number of potential input variables available. In this study, four different feature extraction methods were used to reduce the input vector to train four networks to estimate temperature at different atmospheric levels. The four techniques used were: genetic algorithms (GA), coefficient of determination (CoD), mutual information (MI) and simple neural analysis (SNA). The results demonstrate that of the four methods used for this data set, mutual information and simple neural analysis can generate networks that have a smaller input parameter set, while still maintaining a high degree of accuracy.

A support method for qualitative simulation-based learning system

In this paper, we mainly present a support method of our proposed e-learning system. We employ qualitative simulations because this lets the learners understand the conceptual principles in economic dynamics. First, we define some qualitative values employed on simulation graph model that consists of nodes and arcs. Then, we show the support method using our learning system based on qualitative simulation.

An approach to avoiding shill bids based on combinatorial auction in volume discount

Volume discount is seen as an effective form of electronic commerce and a promising field for applying agent technologies. In current volume discount mechanisms, items are not allocated efficiently to buyers. Namely, social surplus is not maximum in existing volume discount schemes. To solve this problem, we propose a volume discount mechanism based on the sellers reservation price and the payment adjustment value. First, a seller registers his/her items with the evaluation value functions. The sellers evaluation value is sealed and each buyer bids his/her evaluation value as sealed bid. After the deadline, the mechanism determines the allocation of bundles of items. A tentative price is decided and the payment adjustment value is calculated. Finally, the payment amount is calculated. Our mechanism has some key advantages. First, the mechanism is Pareto efficient. Second, our mechanism is a strategy-proof mechanism, that is, it has the incentive compatibility. Third, our mechanism provides individual rationality. Fourth, our mechanism is made based on the volume discount system, where the seller can give a signal indicating a discount for buyers. Finally, our mechanism weakens the influence of false name bids.

A location information system based on real-time probabilistic position inference

One of the difficulties of using Artificial Neural Networks (ANNs) to estimate atmospheric temperature is the large number of potential input variables available. In this study, four different feature extraction methods were used to reduce the input vector to train four networks to estimate temperature at different atmospheric levels. The four techniques used were: genetic algorithms (GA), coefficient of determination (CoD), mutual information (MI) and simple neural analysis (SNA). The results demonstrate that of the four methods used for this data set, mutual information and simple neural analysis can generate networks that have a smaller input parameter set, while still maintaining a high degree of accuracy.

An approach to detecting shill-biddable allocations in combinatorial auctions

This paper presents a method for discovering and detecting shill bids in combinatorial auctions. Combinatorial auctions have been studied very widely. The Generalized Vickrey Auction (GVA) is one of the most important combinatorial auctions because it can satisfy the strategy-proof property and Pareto efficiency. As Yokoo et al. pointed out, false-name bids and shill bids pose an emerging problem for auctions, since on the Internet it is easy to establish different e-mail addresses and accounts for auction sites. Yokoo et al. proved that GVA cannot satisfy the false-name-proof property. Moreover, they proved that there is no auction protocol that can satisfy all three of the above major properties. Their approach concentrates on designing new mechanisms. As a new approach against shill-bids, in this paper, we propose a method for finding shill bids with the GVA in order to avoid them. Our algorithm can judge whether there might be a shill bid from the results of the GVAs procedure. However, a straightforward way to detect shill bids requires an exponential amount of computing power because we need to check all possible combinations of bidders. Therefore, in this paper we propose an improved method for finding a shill bidder. The method is based on winning bidders, which can dramatically reduce the computational cost. The results demonstrate that the proposed method successfully reduces the computational cost needed to find shill bids. The contribution of our work is in the integration of the theory and detecting fraud in combinatorial auctions.

A strategy-proof mechanism based on multiple auction support agents

Agent-mediated electronic commerce has recently commanded much attention. Bidding support agents have been studied very extensively. We envision a future in which many people can trade their goods by using a bidding support agent on Internet auctions. In this paper, we formalize a situation in which people are trading their goods on Internet auctions and employing bidding support agents. Then, we prove that people who use a bidding support agent can successively win trades. Also, we prove that the situation in which every people use a bidding support agent can satisfied strategy proofness and Pareto optimality. Further, we present in the situation, unsupported bidders do not make a positive benefit.