Design And Implementation Of Advertising Delivery System Based On Flink Real-time Data

With the rapid development of the Internet,many Internet companies have appeared in the market.In order to make full use of Internet traffic and reduce labor costs,Internet companies tend to use the programmatic deal model to place online advertising on their own media for profit.Programmatic deal is a mechanism to automatically place advertisements.Its core idea is to use the auction model to sell every advertisement display opportunity.The initiator of the auction is ad exchange that represents the interests of the media,and the bidders are many demand side platforms that represent the interests of advertisers.In order to win the advertisement display opportunity,each demand side platform needs to select an advertisement with the highest value from the advertisement library to participate in the auction.The ad exchange finally selects the advertisement of the highest demand side platform for display.The above-mentioned media is an application software operated by an Internet company.The content of advertisements to be displayed by the media is determined by the demand side platform;The advertisers are those who need to place advertising for their products,and the advertisers hand over their advertising to the demand side platform for management.The advertising delivery system designed in this paper realizes the function of the demand side platform in technology.The system uses Flink framework to process and count data.Flink is a distributed processing framework that supports high throughput,low latency and high performance.Flink can calculate massive data in real-time.The system mainly includes five major functional modules: advertising information management module,advertising obtain module,advertising filter module,advertising offer module,and data statistics and analysis module.The system can obtain an advertisement with the highest value through the three modules of advertising obtain module,advertising filter module,advertising offer module.The author participated in the design and implemented the following five functional modules.Advertisement information management module: the module provides advertisers with services for the entry of advertisement information and the viewing of advertisement delivery effects.Except for the module,all other functional modules of the system are automatically completed by the system without personnel involvement.Advertising obtain module: the module constructs query statement through bidding request information and user characteristic information,and obtains advertisement data in database through this statement.When the module constructs the query statement,it does not consider the advertiser's delivery and budget consumption on that day.The system needs to filter out the advertisements that may appear excessive delivery in the advertising filter module.Advertising filter module: the function of the module is to use a series of filtering methods to remove advertisements that may appear to be over-delivered.The design and implementation of the module is the core of the system.Advertising offer module: the module first sorts the remaining advertisements after filtering,then determines the bidding price of each advertisement,and finally sends the highest value advertisement to the ad exchange to complete the bidding response.Data statistics and analysis module: the function of the module is to process and count data of user behavior,and provide data support for the advertising filter module.The results of online operation show that the system can select the highest value advertisement to participate in the auction initiated by the advertising trading platform according to the bid request information.The design and implementation of this system can make full use of the value of data for the company to bring a steady stream of economic income,and at the same time can provide a design scheme for more companies to automatically select advertising.