搜索竞价广告关键词优化问题研究

Study of Search Engine Advertising Keyword Optimization

作者: 专业:计算机应用技术 导师:梁永全 年度:2010 学位:硕士  院校: 山东科技大学

Keywords

search engine advertising, keyword optimization, click-through rate, concept hierarchy, language pattern

        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竟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的生成的关键词。
    Currently, search engine advertising is the major delivering manner and the effective marketing method of network advertising, advertisers exhibit their services and products in order to obtain economic benefits through their delivered advertisements, meanwhile the search engine users read the advertisements and advertising information through the matching between the search keywords they input and the bidding advertising keywords they purchase. Whether the ads can be discovered accurately by users who are interested in them and obtain more chances to exhibit to users depends on the keyword optimization for search engine advertising. At present, a common demand for advertiser is to obtain plenty of keywords automatically which are related to a specific advertisement and can bring more economic benefits. The related issues corresponding to this demand is search engine advertising keywords optimization. Search engine advertising keywords optimization is a hot topic and a difficult point in the research area of search engine advertising, its difficulty is how to generate a group of bidding keywords which contains a mass of keywords which are related to the advertisements delivered by advertisers and can help advertisers obtain more economic benefits.In order to solve the difficulty existing in the area of Search engine advertising keyword optimization, we argue that advertising keyword optimization can be divided into three stages to process. The first stage is advertising keyword extraction. The main task of this stage is to design the model of advertising keyword extraction and extract keywords as the seed keywords from advertisement. We apply an extraction model based on language pattern mining. This model can guarantee that the seed keywords have a very high relevance with advertisement. The second stage is advertising keyword expansion. The main task of this stage is to design the model of advertising keyword expansion and generate plenty of candidate bidding keywords which are related to seed keywords. We apply a expansion model based on concept hierarchy, this model can guarantee the large number of the candidate bidding keywords and the high relevance with the seed keywords. The third stage is candidate bidding keywords optimization and selection. The main task of this stage is to design the model of optimization to select more superior keywords from candidate bidding keywords. We apply a optimization model based on click-through rate prediction. This model can guarantee that the optimal results can bring greater economic benefits for the advertisers.Based on the above works, we make the experiment to evaluate the effect of the search engine advertising keyword optimization method which is composed of the above three models. Firstly, we verify that the keyword extraction model based on language pattern mining is better than traditional keyword extraction model. Secondly, we verify that click-through rate optimization model base on LRM has a high precision on experimental result. These two experimental results provide very big support to the effect evaluation of entire optimization algorithm. Finally, we implement an experiment to compare the search engine advertising keyword optimization method with leading advertising keyword suggestion tools. The result shows that our keyword optimization method is better than leading advertising keyword suggestion tools.
        

搜索竞价广告关键词优化问题研究

摘要5-6
Abstract6-7
1 引言10-14
    1.1 本文的研究背景10-11
    1.2 本文的研究内容及意义11-12
    1.3 本文的工作12-13
    1.4 本文的组织13-14
2 相关研究工作14-22
    2.1 搜索竞价广告关键词优化与关键词抽取14-15
    2.2 搜索竞价广告关键词优化与查询扩展15-18
    2.3 搜索竞价广告关键词优化研究现状18-20
    2.4 小结20-22
3 种子关键词生成22-29
    3.1 问题分析22-23
    3.2 模型简介23-24
    3.3 语言模式挖掘24-28
    3.4 种子关键词抽取28
    3.5 小结28-29
4 基于语义模型的候选关键词集的生成29-39
    4.1 语义模型概述29-31
    4.2 概念层次结构模型构建31-36
    4.3 候选词集生成36-38
    4.4 小结38-39
5 基于点击率预测的关键词优化算法39-49
    5.1 影响广告收益的经济因子分析39-40
    5.2 广告点击率预测模型40-46
    5.3 基于关键词点击率预测的优化模型46-47
    5.4 小结47-49
6 实验与分析49-61
    6.1 搜索竞价广告关键词抽取实验49-54
    6.2 广告点击率预测实验54-58
    6.3 广告关键词优化综合实验58-60
    6.4 小结60-61
7 总结与展望61-63
致谢63-64
攻读硕士期间主要成果64-65
参考文献65-68
        


本文地址:

上一篇:基于本体的物流信息集成系统的研究
下一篇:基于数据挖掘的客户行为分析和预测研究

分享到: 分享搜索竞价广告关键词优化问题研究到腾讯微博           收藏
评论排行
公告