关键词广告中的最优广告策略研究

Keyword advertising in internet achieves great rewards recently. However, the theoretical study does not keep up with the development in market. Due to commercial confidentiality reasons, the engine supplier does not reveal keyword ad ranking system in details and does not provide the whole bidding data. Advertisers are difficult to obtain competitors' bidding data. Therefore, the progress of relevant research is slow. The applicability of existing research is limited because of its simplistic assumptions in game model or empirical model.In addition, the study of advertising strategy in some areas is still empty. In this research, bayesian inference, optimal control theory, dynamic game theory, heuristic algorithm,empirical analysis and numerical experiment are employed to study bidding strategies based on existing results. The main contents include ： 1. Keyword advertising comprehensive effectiveness evaluation; 2.Advertising value evaluation of keyword and keywords selection strategy; 3. The budget dynamic allocation strategy for multi-stage in multi-keyword bidding in single or multi search engine, develop effective algorithm for solution; 4. Game model of bidding strategy under complex conditions in single or multiple engine platforms. This project aims to reveal various factor that affecting keyword advertising, to provide scientific theory basis for the implementation of keyword advertising business activities. Furthermore, this project makes effort to improve and expand theory of internet marketing.

目前互联网关键词广告商业模式已取得巨大成功，但理论研究却相对滞后。出于商业保护的目的，引擎商不愿公开关键词广告机制细节和整体广告数据，广告商也无法获取竞争对手的广告数据，相关的研究因此进展缓慢。已有的广告策略研究大都基于简化的博弈模型分析或实证模型分析，其适用性受到限制。此外，关键词广告策略某些领域的研究几乎还处于空白。 本项目在已有研究成果基础上，运用贝叶斯推理、最优控制理论、动态博弈、现代启发式算法、实证分析、数值实验和仿真等方法围绕关键词广告最优策略展开系列研究，主要内容包括：1.关键词广告效果综合评估；2.关键词广告价值评估及选择策略；3.单引擎平台和多引擎平台中多阶段多关键词广告预算动态分配策略；4.复杂条件下单引擎平台和多引擎平台中竞价博弈模型及竞价策略研究。本项目旨在揭示影响关键词广告策略的各种相关因素，为关键词广告活动的有效实施提供科学依据，完善网络营销理论体系。

在互联网关键词广告商业模式已取得巨大成功而理论研究却相对滞后的背景下，本项目运用统计学理论、最优控制理论、动态博弈、现代启发式算法、实证分析、数值实验和仿真等方法围绕关键词广告最优策略展开系列研究，包括：1. 构建了基于层次贝叶斯方法的转化率模型和基于LASSO变量选择方法的广告点击率预测模型，以及基于转化率和点击率为主要衡量指标的广告效果评价体系，并研究了影响这些指标的相关因素。2.研究了关键词特性对广告效果的影响，在此基础上制定关键词的选择策略。3. 研究并构建了单引擎广告时序预算分配模型和多个搜索引擎上同时投放广告的动态预算分配模型，并确立最优分配策略。4. 建立了完全信息和不完全信息下的双关键词竞价模型，提出了该模型下的最优竞价策略。.以上的研究成果为广告商的关键词广告实施提供一定的理论依据，并在一定程度上完善了网络营销理论体系。