搜索引擎广告中广告商状态建模

搜索引擎广告是目前互联网上一种非常成功的商业模式，它已成为搜索引擎公司的主要收入来源并为广告商们提供了许多商机.搜索引擎、广告商和搜索用户构成了搜索引擎广告的3个主要组成部分——搜索引擎提供技术和服务、广告商提供广告内容、用户浏览并点击广告.其中搜索引擎相关技术以及用户行为都有比较多的研究和成型的技术，但对广告商尤其是广告商状态的研究却并不多见.基于此背景，对搜索引擎广告中广告商的状态进行了深入的研究.在方法上按照广告商相关广告的展示次数、点击次数以及广告费用来对其潜在状态进行描述和划分，并使用隐Markov模型对广告商的时序状态进行建模.重点在于将机器学习和数据挖掘的方法应用于广告商的建模之中，并取得了不错的预测正确率.

Advertiser Status Modeling in Sponsored Search

Sponsored search is a successful business model currently on the Internet. It has become the main income source of search engine companies and has offered great opportunities for advertisers. Search engine, advertiser and user are the three main components in sponsored search. Search engine offers technology and advertisement service, advertiser offers advertisement contents while user views and clicks advertisements. Related technologies of search engine and user behavior have been studied and developed by many researchers. However, advertisers especially the status of advertisers has not been studied well in the area. Based on such situation, we conducted deep analysis on the status of advertisers. Technically, such status is demonstrated by the impression number, the click number and the cost of the advertiser’s advertisements. Then hidden Markov model is utilized in demonstrating such sequential status of advertisers. The focus is on introducing methodologies of machine learning and data mining into advertiser modeling. At the same time, we get high precision when predicting advertiser status with our model, which convinces us that such method is suitable.