Click-through rate estimation method, device, computing apparatus, and storage medium

Abstract

The invention provides a click-through rate estimation method, a click-through rate estimation device, a computing apparatus, and a storage medium. The method includes the following steps that: a feature evaluation network with user features and item features adopted as nodes is established; a click-through rate estimation model is established based on the feature evaluation network; and the click-through rate of users for an item can be estimated based on the click-through rate estimation model. Thus, the click-through rate estimation model can reflect a relationship between the user featuresand the item features in the form of a network, so that the click-through rate of an advertisement can be easily estimated.

Description

technical field [0001] The invention relates to the technical field of Internet applications, in particular to a click rate estimation method and device. Background technique [0002] Click-through rate estimation has a wide range of applications in computational advertising, recommendation systems, and other fields. By making statistics on the click-through rate of advertisements, it is possible to understand the advertisements that different users are interested in, so as to push advertisements to each user more accurately, so as to increase the click-through rate of advertisements. Accurate advertising click-through rate estimation can improve the real advertising click-through rate, thereby increasing advertising revenue. [0003] According to different advertising data characteristics, different advertising click-through rate estimation models can be used to obtain better estimation results. Advertisement click-through rate prediction models such as logistic regressio...

AI Technical Summary

Problems solved by technology

[0004] The current click-through rate estimation method uses artificial feature engineering combined with logistic regression algorithm, first extracts the relevant features of the keywords input by the user and candidate advertisements, and inputs the relevant features into the linear model to obtain the click-through rate of the corresponding advertisement, but the extraction of relevant features High manpower and time costs are required

Moreover, this method is too subjective and easy to screen out a large number of useless features to affect the accuracy of model evaluation, and the logistic regression algorithm is a generalized linear model, the linear model has limited learning ability, and the relationship between each feature in the model and the prediction result is independent of each other. are not affected by each other, and thus cannot learn nonlinear relationships between features, resulting in less accurate predictions obtained

In addition, the linear model will cause other problems, such as how to select features in high-dimensional scenes, generally using artificial feature engineering, as the process deepens, the benefits of this method will gradually reach the upper limit

Images