Advertisement putting method and system

Abstract

The invention discloses an advertisement putting method. The method comprises the following steps of: obtaining click rate feature parameters of a predetermined history put advertisement in a predetermined history time period; crossly combining at least two types of feature parameters in the feature parameters to obtain a combined feature parameter, and digitalizing the feature parameters and thecombined feature parameter to obtain a feature parameter set; carrying out feature selection on the feature parameter set to obtain a target feature parameter; and training a prediction model on the basis of the target feature parameter, and carrying out prediction calculation on a click rate of a to-be-predicted advertisement through the prediction model. The invention furthermore provides an advertisement putting system. According to the method and system, in the advertisement click rate prediction process, the feature parameters are combined and digitalized, and then feature selection is carried out on the feature parameters to improve the feature selection quality, so that the condition of mistakenly deleting features of all the information amounts can be avoided to the greatest extent, the calculation pressure and response time are decreased, and the timeliness and click rate prediction precision of advertisement putting are ensured.

Description

technical field [0001] The invention relates to the technical field of Internet advertisements, in particular to an advertisement delivery method and system. Background technique [0002] At present, in the Internet field, click-through rate prediction is an important part of calculating the cost of advertising. For example, click-through rate forecasting can be applied to push personalized advertisements to users, thereby increasing the overall click-through rate and conversion rate of advertisements, thereby increasing the value of advertisements. Secondly, the click-through rate prediction can also be used to estimate the price of exposure, so that advertisers can bid for real-time bidding or exposure in the advertising network, so as to buy the exposure they need at an appropriate price. [0003] The common click-through rate prediction methods in the industry are as follows: the first average click-through rate prediction, that is, the average click-through rate of a ce...

AI Technical Summary

Problems solved by technology

[0005] The average CTR prediction and the CTR prediction by group are too rough for the crowd and cannot be applied to more features, so the CTR prediction accuracy of these two methods is low

The model-based click-through rate prediction considers too many characteristics of advertisement placement, and may add some features that are not informative, which will affect the accuracy of the overall click-through rate prediction, and will also lead to excessive calculation pressure on the advertisement delivery server, long response time, and reduced It affects the speed of advertising delivery and affects the real-time performance of advertising delivery.

Images