面向单一搜索引擎的关键字广告竞价策略研究

发布时间：2018-05-03 11:10

本文选题：贝叶斯网络 + 竞价策略　；参考：《哈尔滨工业大学》2013年硕士论文

【摘要】：自搜索引擎问世以来，就以其锐不可当的势头风靡全球。究其原因是互联网上的信息增长速度过快，各行各业的信息都杂糅在一起，致使用户在搜索自己所需的信息时变得无从下手。因此可以帮助人们有效实现精准搜索的搜索引擎成为互联网用户必不可少的网络工具。搜索引擎服务提供商与各广告商也正是看到这一潜在商机，使得搜索引擎上的关键字广告业务迅速发展起来，关键字广告也已成为各企业不可或缺的一种电子商务营销模式。但是在关键字竞价过程中，并不是广告商出价越高就能获得更好的广告排名和更高的利润。关键字广告竞价一直处于动态变化中，如何有效地在搜索引擎上投放关键字广告是各广告商所关注的问题，也是本文研究内容的重点。本文在网上拍卖理论的基础上引出了关键字广告竞价拍卖的相关机制，，并利用贝叶斯网络预测模型为广告商提供竞价策略。首先本文根据关键字广告竞拍过程提取出模型所涉及的变量，并在关键字广告竞拍过程和对变量关系的分析基础上构建了贝叶斯网络结构模型，然后利用广告商的关键字广告投放历史数据确定贝叶斯网络的参数模型。本文依据上一期的广告点击量对当前广告排名是否有影响，分别构建了静态贝叶斯网络预测模型和动态贝叶斯网络预测模型。实验结果表明，动态贝叶斯网络比静态贝叶斯网络的预测效果更精确。同时，本文构建了基于贝叶斯网络的决策模型，可以求出无预算约束条件下广告主获取最大利润时的具体出价，并依据这个模型为广告主提供存在预算约束时的竞价策略。
[Abstract]:Since the advent of search engines, with its irresistible momentum swept the world. The reason is that the information on the Internet is growing too fast, and the information of various industries is mixed together, which makes it impossible for users to search for the information they need. Therefore, search engine which can help people to realize precision search has become an indispensable network tool for Internet users. Search engine service providers and advertisers also see this potential business opportunities, making the search engine keyword advertising business has developed rapidly, keyword advertising has become an indispensable e-commerce marketing model. But in the process of keyword bidding, it is not the advertisers who bid higher to achieve better advertising ranking and higher profits. Keyword advertising bidding has been in a dynamic change, how to effectively place keyword ads on search engines is a problem that advertisers are concerned about, and also the focus of this paper. Based on the theory of online auction, this paper introduces the related mechanism of keyword advertising auction, and provides the bidding strategy for advertisers by using Bayesian network prediction model. Firstly, this paper extracts the variables involved in the model according to the bidding process of keyword advertising, and constructs a Bayesian network model based on the analysis of the bidding process and the relationship between the variables. Then, the parameter model of Bayesian network is determined by using the keyword advertising history data of advertisers. In this paper, according to whether the number of advertising hits in the previous issue has an effect on the current advertising ranking, the static Bayesian network prediction model and the dynamic Bayesian network prediction model are constructed. Experimental results show that dynamic Bayesian networks are more accurate than static Bayesian networks. At the same time, this paper constructs a decision model based on Bayesian network, which can find out the concrete bid of the advertiser when he gets the maximum profit under the condition of no budget constraint, and provide the bidding strategy for the advertiser with budget constraint according to this model.
【学位授予单位】：哈尔滨工业大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：F713.8;F224

【相似文献】