联系我们

论文查询结果	返回搜索

论文编号：	1555
作者编号：	1120060713
上传时间：	2009/6/18 11:35:25
中文题目：	基于网上行为特征的网络精准广告
英文题目：	Research of Online Targeted Ad
指导老师：	严建援,林漳希
中文关键字：	网络精准广告;网上行为;广告投放
英文关键字：	Online targeted advertising;On
中文摘要：	在现代社会中，由于需求的多元异质性越来越突出，消费者的需求也越来越分化，网络精准营销相对于“广而告之”的传统网络营销手段来说，在成本和效率上的优势日益凸现。不仅如此，激烈的竞争也使现代企业需要一种更精准、可衡量且投资回报高的营销手段。网络精准广告是商务智能在网络广告营销领域的具体应用，其基本思想就是运用数据分析和数据挖掘技术从海量的网络用户浏览历史数据中挖掘分析出有价值的信息和规律，以辅助网络广告投放决策。其最终目的就是在正确的时间将最合适的网络广告投放给最合适的网络用户。随着电子商务的发展，互联网广告点击率呈现了不断下降的趋势，而网络精准广告提高了网络广告的点击率，避免了网络营销资源的浪费，使网络广告更为精准，有效。网络精准广告本身也已经成为了网络广告的新趋势。 本文以网络精准广告这一商业实践现象为研究基础，在企业调研和访谈的基础上，从理论和技术两个层面深入挖掘了网络精准广告中的研究问题。具体而言，论文的主要研究工作如下： 第一，在理论层面上验证网络用户网上行为特征与网络广告点击之间的关联关系。网络用户网上行为特征与网络广告点击之间的关联关系是网络精准广告服务的基本假设。通过对实践数据的统计分析，本文证实了运用网络用户网上行为特征数据来辅助决策网络广告的投放是有意义的，通过使用数据挖掘等技术来对网络用户网上行为的历史数据进行分析，能够挖掘出非常有价值的信息和规律，从而让我们能够筛选出最合适的网络广告来投放给访问目标网页的网络用户。 第二，网络精准广告的实施不是一两个数据挖掘模型就可以解决的问题，它是一个系统的问题，一个由若干功能模块相互协作才能完成的工作，包括广告的组织和存储，分类决策模型，网络用户网上浏览实施数据的获取，网络广告的投放分配机制等等。本文在实施层面上提出了网络精准广告实施的系统框架，系统框架包括网络用户，广告主，网络广告发布商和网络精准广告服务商四个主体，并将系统的功能划分为三个部分，即广告管理，广告决策优化模型和广告实时投放。 第三，网络广告精准技术是网络精准广告服务的核心，也是各网络精准广告服务商获得竞争优势的根本。本文在广告精准技术方面提出了一种网络广告聚类加分类决策模型的解决方案，并使用来自于实践的网络用户网上行为数据针对分类决策模型设计了一个实验，分析和比较各种分类方法。 第四，在已有研究的基础之上提出一种基于广告排序的网络精准广告投放分配机制。广告投放分配机制所解决的问题是在各种资源限定的条件下，将网络广告投放机会分配给各网络广告。广告排序的思想能够让我们在追求收益最大化的同时兼顾分配的公平性，这使得基于广告排序的网络精准广告投放分配机制能够更好的兼顾短期和长期利益。为了验证基于广告排序的投放分配机制的可行性和有效性，本文还运用数学建模仿真的方法，仿真了三种网络精准广告投放分配机制的分配投放过程，并从三种投放分配机制的仿真过程中，得到了一些反映网络精准广告投放分配过程中的一些推论。
英文摘要：	In modern society, consumers’ demands are becoming more and more various because of demand’s diversification. Comparing with the “Speaking Generally” traditional online advertising, online targeted advertising is becoming more and more advantageous, especially in cost and efficient. Intense competition makes firms require a more targeted and measurable marketing method with high revenue of investments. Online targeted advertising is the application of business intelligence in online advertising area, and its basic idea is using data mining and data analysis technique to find the useful information and rules from huge amount of web user’s surfing data and use them to support online advertising decisions. The ultimate objective of online targeted advertising is to post suitable online advertisement to the suitable web user in the right time. As E-commerce is developing fast, the click through rate of online advertisement keeps dropping. Online advertising helps people to find a way to raise their online ads’ click through rate. This avoids waste of online marketing resource and makes online advertisement more targeted and more efficient. Online targeted advertising itself has already become a new development direction of online advertising. This paper is based on the practical commercial phenomenon, online targeted advertising. After the interview and discussion with the researchers and managers from industry, this research is studying several research issues from both theoretical and technical perspectives. In detail, the main works of this research are shown as follows: Firstly, the paper tests the association relationship between web users’ online behavior characters and the clicks of online advertisement. The association between web users’ online behavior characters and the clicks of online advertisement is the basic assumption of online targeted advertising. Through the statistical analysis of practical data, this paper verifies that web users’ online behavior data is meaningful to support online advertising decisions. By using data mining technique to analyze web user’s online behavior data, we can find lots of valuable information and rules and these can let us post the suitable online advertisement to the suitable web user that is visiting targeted web page. Secondly, the implementation of online targeted advertising is definitely not a simple problem that can be figured out by using one or two data mining models. It is a systematic problem, a cooperating work that contains several functions, including advertisement organizing and store, classification model, real-time session acquirement, online advertisement allocation and so on. This paper proposes an online targeted advertising system framework, which include four subjects: web user, advertiser, publisher and online targeted advertising service provider, and the framework divides system function into three part: advertisement management, optimum decision model and real-time advertisement publishing. Thirdly, targeting technique is the core of online targeted advertising service, and also the base of online targeted advertising service provider’s competitive advantage. This paper proposes a new solution for targeted advertising, online advertisements clustering plus classification model. Also the paper designs an experiment by using the practical data to analyze and compare several classification models and tries to find the most suitable way for online targeted advertising. Fourthly, based on the existed research, this paper proposes an online targeted advertising allocation mechanism based on advertisement ranking. The problem that advertisement allocation mechanism can solve is allocating impressions to online advertisements in the condition of the limited resources. Advertisement ranking can let us consider the uniform of advertisement allocation while we are pursuing maximum profit. Also it makes online targeted advertisement allocation be more profitable in both short time period and long time period. In order to test the feasibility and efficiency of advertisement ranking based mechanism, we use mathematical model to simulate the allocating process of three online targeted advertising allocation mechanisms, and proposes some findings.
查看全文：	预览  下载(下载需要进行登录)