The Design And Implementation Of Advertising System Based On RTB Mode

With the rapid development of the Internet,online advertising came into being.Compared with traditional media advertising and highly favored outdoor advertising,online advertising is becoming one of the most profitable business models of Internet companies,showing huge market potential.Internet advertising has many different manifestations which is not only limited to Banner advertising on the web pages but also includes search engine keyword advertising,search rank etc.However,such many advertising models inevitably have high cost,ineffective audience reach and other issues,leading to unsatisfactory effect.Thus,a new type of real-time bidding advertising model recently appeared and developed rapidly.By evaluating for each user's behaviors and auction,RTB mode avoids the disadvantages of the old stylized launch.For this mode,we cooperate with a number of Ad Exchange platform to expand advertising flow and improve advertising effectiveness.This paper implements an advertising system supporting the RTB mode on the basis of the original advertising search engine.Firstly,this paper describes the main works done by the advertising system.It analyzes requests from Ad Exchange platform and the search engine choose the satisfied launch and creative according to predicted CTR of every launch,bidding price after a series of filtering.Meanwhile,this paper introduces the specific implementation work of advertising system from requirements analysis,design and implementation.The whole system based on Spring,Struts2 and MyBatis framework,combined with thread pool,recurring tasks,aspect oriented programming and Hive.In CTR estimating module,we use logistic regression models to classify and train feature data by Spark frame.And for the bid module,we use Storm framework to pull win notices and and calculates new bidding price.The advertising system introduced by this paper solves problems mentioned above,achieves automatic sales,cuts costs and improves advertising accuracy and ROI.