基于信息增益比率分类的多Agent序贯拍卖模型  

A MULTI-AGENT SEQUENTIAL AUCTION MODEL BASED ON THE RATIO OF INFORMATION GAIN

作　　者：赵月[1] 

机构地区：[1]安阳师范学院,河南安阳455001

出　　处：《南阳理工学院学报》2011年第6期19-23,共5页Journal of Nanyang Institute of Technology

摘　　要：大多数序贯拍卖模型中的标的只具有一个属性,本文构造了一个标的兼有共同价值和私人价值两个属性的序贯拍卖模型。考虑在拍卖过程中公开标的信息,分别给出此模型下买家Agent在第一价格密封拍卖规则(The First Price Sealed Auction,FPSA)和第二价格密封拍卖规则(The Second Price Sealed Auction,SPSA)中的竞价策略函数。提出一个基于标的信息增益比率的分类算法LIWNB,使得买家Agent在提交竞价时能准确分类当前拍卖信息,精确估计标的的共同价值。实验结果显示,算法LIWNB在一定条件下具有较高的分类性能。In most literatures the object for sale has one feature.An sequential auction model whose object for sale has common value and private value is proposed.Revealed information is involved with auction model in this papers.It has been studied that buyers/sellers and equilibrium strategies will be impacted when the auctioneer reveals information about the object for sale in FPSA or SPSA.On the basis of the ratio of information gain,a local information weighted naive Bayes classifier is given to the buyer agents to learn and classify information in auctions.Some positive results are obtained in experiments.Success in learning and classsifying revealed information shows the effectiveness and valuableness of this algorithm.

关 键 词：多Agent序贯拍卖 共同价值 私有价值 信息增益比率 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]