一种基于多Agent系统的在线广告竞价模型  

AN ONLINE ADVERTISING BIDDING MODEL BASED ON MULTI-AGENT SYSTEM

作　　者：张文明[1] 王小平[1] 

机构地区：[1]同济大学电子与信息工程学院,上海200092

出　　处：《计算机应用与软件》2017年第2期133-141,共9页Computer Applications and Software

摘　　要：针对在线广告竞价这一具体应用场景,本文在价高者得的传统竞价原则上,结合竞价双方的相似度阈值控制,提出一种适用于在线广告交易的竞价模型,即SMOBM(the Similarity-based Multi-agent Online-advertising Bidding Model)模型。该模型使用多Agent系统建模,根据市场角色设计相应Agent,并设计出需求聚合Agent组织和资源聚合Agent组织。买卖双方使用回合制的拍卖协商方法进行连续叫价,并在系统全局时钟的控制下产生交易结果。最后,使用Repast Simphony仿真软件对模型进行实现,将基于组织模型和不基于组织模型这两种情况进行对比,结果表明在选择合适的相似度阈值时,基于组织模型的交易成功率高,且在存在恶性竞争时整体交易表现出较好的鲁棒性。Since the online advertising auction is becoming a new application scene,the SMOBM model based on the traditional principle of success of highest-price-offer and two sides of the transaction similarity threshold controlhave been proposed in this paper. This model uses Multi-Agent system to design,according to the market role,the corresponding Agent and Agent organization are designed, including the demands aggregation Agent organization and resources aggregation Agent organization,the buyers and sales uses a round-based negotiation method to bid for the auction,and produce the results of transactions under the control of global clock. The paper used the Repast Simphony simulation software to realize the model,the model is compared in two cases,one used the organization and the other didn＇t. In the choice of a suitable similarity threshold,the results show that the former has higher transaction success rate and overall transaction showed robust in the existence of malicious competition.

关 键 词：多AGENT系统 在线广告竞价 AGENT组织 相似度 

分 类 号：TP399[自动化与计算机技术—计算机应用技术]