Study On Bidding Behavior Forecasting Model Of Advertisers On HMM

Search engine is the most important information searching platform,keyword advertising is its main source of economic income.To search engines,to maximize their own profit is the main starting point and foothold.And advertisers as the only payer of search engine,it is one of the most important part in keyword advertising auction.Therefore,whether can accurately predict advertisers’ future bidding sequence is the key factors that affecting the auction mechanism to maximize the revenue of the search engine.The main work of this paper is predicting the future bidding behavior by formulating advertisers’ bidding behavior model.Firstly,establish HMM advertisers bidding behavior prediction model by analyzing and studying the bidding behavior of advertisers;then obtaining from the relevant historical data from the keyword advertising’s simulation game platform(TAC/AA),and training this model in machine learning,obtaining the prediction model based on HMM advertisers bidding behavior finally.Finally,in order to optimize this model,we considering the rational of advertisers’ bidding.We can extract feature vectors which related to the rational of advertisers’ bidding from the historical bidding data,so as to get advertisers’ rational bidding behavior.Considering advertisers’ rational will make the bidding behavior model more precise,and it also could make a greater contribution to the formulation of search engine keyword advertising auction mechanism.