搜索竞价广告关键词优化算法与实验

发布时间：2018-04-30 12:01

本文选题：搜索竞价广告 + 关键词优化　；参考：《电子科技大学》2011年硕士论文

【摘要】：搜索竞价广告是当前互联网提供的主要的网络广告投放方式和最有效的营销手段,广告主通过投放的广告向用户展示服务和产品以获得经济收益,而搜索引擎用户则通过输入的查询关键词与广告竞价关键词的匹配来查询广告并查看广告信息。搜索竞价广告关键词优化对于广告能否准确的被用户定位并获得更大的展示机会有着至关重要的作用。目前广告主的一个普遍需求是自动获得大量跟广告相关的且能够带来最大收益的竞价关键词以提高广告的展示机会和转化几率。这个需求对应的相关问题即搜索竞价广告关键词优化问题。搜索竞价广告关键词优化是当今搜索竞价广告领域的研究热点和难点,它的难点在于如何为广告生成大量的、相关的并能获得较高经济效益的竞价关键词。针对目前搜索竞价广告关键词优化领域存在的问题,本文提出将广告关键词优化分为三个阶段进行处理。第一阶段,广告关键词抽取阶段。这一阶段的主要目标是根据搜索竞价广告的特点进行广告关键词抽取模型的设计并抽取广告中的关键词作为种子关键词。本文使用基于语言模式挖掘的抽取模型,这种模型能保证种子关键词与广告具有很高的相关性。第二阶段,种子关键词扩展阶段。这一阶段的主要目标是依据种子关键词设计广告关键词扩展模型,以扩展出大量的与种子关键词相关的候选竞价关键词集合。本文使用基于概念结构的扩展模型,这种模型能保证生成的关键词数量众多并且与种子关键词相关度较高。第三阶段,候选竞价关键词优化选择阶段。这一阶段的主要目标是设计优化模型对候选竞价关键词集合进行优化选择。本文使用基于点击率预测的优化模型,这种模型能保证优化结果能够为广告主带来更大的经济收益。在上述工作的基础上,本文用实验验证了由上述三种模型组成的搜索竞价广告优化方法的有效性。首先验证了基于语言模式挖掘的关键词抽取算法在广告关键词抽取中优于传统的关键词抽取算法。然后验证了基于LRM的点击率优化算法也具有较高的准确率。这两个实验结果对整个优化算法的有效性验证起到极强的支持作用。最后将搜索竞价广告优化方法与主流广告关键词推荐工具进行了比较实验,实验结果显示,本文的搜索竞价广告优化方法生成的竞价关键词优于主流广告关键词推荐工具生成的关键词。
[Abstract]:Search auction advertising is the main online advertising mode and the most effective marketing means provided by the Internet at present. Advertisers display services and products to users through the advertisements they put in in order to obtain economic benefits. Search engine users query advertisements and view advertising information by matching the input keywords with the keywords of advertisement bidding. Search auction advertising keyword optimization can be accurately targeted by the user and obtain greater opportunities for display has a vital role. At present, a general demand of advertisers is to automatically obtain a large number of advertising related bidding keywords that can bring the maximum revenue to improve the chances of advertising display and transformation. The related problem of this requirement is the optimization problem of search bid advertisement keyword. Search auction advertising keyword optimization is the research hotspot and difficulty in the field of search auction advertising. The difficulty lies in how to generate a large number of relevant and high economic bidding keywords for advertising. Aiming at the problems existing in the field of keyword optimization in search bid advertising, this paper proposes to divide the optimization of advertisement keywords into three stages. The first stage, advertising keyword extraction stage. The main goal of this stage is to design a keyword extraction model according to the characteristics of search advertising and extract keywords as seed keywords. In this paper, the extraction model based on language pattern mining is used, which can ensure the high correlation between seed keywords and advertising. The second stage, seed keyword expansion stage. The main goal of this stage is to design an advertising keyword extension model based on seed keywords to expand a large number of candidate bidding keyword sets related to seed keywords. In this paper, we use an extended conceptual structure model, which can guarantee a large number of generated keywords and high correlation with seed keywords. The third stage, candidate bidding keyword optimization selection stage. The main goal of this stage is to design an optimization model to optimize the selection of candidate bidding keyword sets. In this paper, an optimization model based on the prediction of click rate is used, which can ensure that the optimization results can bring greater economic benefits to advertisers. On the basis of the above work, the effectiveness of the search bidding advertising optimization method composed of the above three models is verified by experiments in this paper. Firstly, it is verified that the keyword extraction algorithm based on language pattern mining is superior to the traditional keyword extraction algorithm in advertising keyword extraction. Then it is verified that the LRM-based click rate optimization algorithm also has a high accuracy. These two experimental results support the validity of the whole optimization algorithm. Finally, the optimization method of search bid advertisement is compared with the mainstream advertising keyword recommendation tool. The experimental results show that, The keywords generated by the search auction advertising optimization method are superior to those generated by the mainstream advertising keyword recommendation tool.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2011
【分类号】：TP391.3

【参考文献】