Research Of New Media Advertising Online’s Revenue Management Based On Big Data

Traditional internet adverting is based on the Guaranteed-Delivery. But as the online advertising market shifts to the buyer’s market, only depending on the Guaranteed-Delivery to cover all the advertising display opportunities is becoming more and more difficult. Owing to the perishability of the advertising resources, media can’t get profits from these unemployed advertising display opportunities, which will lead to the waste of resources. In the condition that advertising position and page display keeping a certain quantity, advertising inventory is constant. So as to one advertising display, if it can’t get profits as much as possible by reasonable pricing mechanism, it is also the waste of resources. With the new media’s page display of personal computer end is growing slowly or even at a standstill, how to maintain the increase of profits is becoming a problem demanding prompt solution.Firstly, this thesis studies the capacity control strategy that based on audience orientation. Through the media access log for big data analysis, to realize the browser user-directed and user interest population orientation, and then directed to the audience the results as a basis for capacity control, so as to increase profitability of media party. In this thesis, using a questionnaire to acquire the part of the user population attribute information firstly, and then analyze the whole amount of users browse the log pages and videos with big data, using Naive Bayesian model to predict their population attribute information. The thesis also studies the implementation of user interest orientation, through the page access log for big data analysis, based on the user browse the page, to infer their point of interest. The different advertising display that brought though the division of demographic attributes and user interest groups, can be sold to specific advertisers, and achieve capacity control.Secondly, in terms of advertising pricing strategy, this thesis studies the location-based real-time auction bidding strategy. By studying the characteristics of online advertising, the thesis designed a set of online advertising auction mechanism, and has proved that use the generalized second price as search strategy, can make continuous auction reach Bayes-Nash equilibria, so that the media may obtain a maximum income. In addition, This thesis also studied that based on the past for a period of time display advertising and click log for big data analysis, to achieve ad click through rate forecast before, thus to achieve the CPC and CPM bid advertisers bidding under the same standard.Finally, this thesis applies the research results mentioned above to the online advert ising business of a certain Internet media and make an empirical analysis about its sci entificity. The result shows that, after the media use the audience orientation strategy t o make volume control, the higher label strength the user with, the more easy the user will click the advertisements related to his or her population attributes or interests. A nd as to the advertisements which is auctioned off based on the CTR prediction model , their click-through rate is 12% higher than the advertisements which are not make la unching decision based on the CTR prediction model.This thesis focus on the new media online advertising’s revenue management methods based on big data, which can provide theoretical support for the strategy of internet media optimizing display advertising revenue. Using this method can promote the advertiser, the media and the audience to achieve win-win to the maximum extent. Furthermore, this thesis can also elicit further study on the computational advertising and revenue management.