Method and system for obtaining advertisement click-through rate pre-estimation model

Abstract

The invention discloses a method and system for obtaining an advertisement click-through rate pre-estimation model. The method comprises the steps of: obtaining historical click journal data generated according to data of users, query key words and clicked advertisements in query results and the frequency of advertisement exhibition; respectively performing clustering dimension reduction on the users, the query key words and the advertisement data according to the frequency of the advertisement exhibition and respectively obtaining the user data after clustering, the query key word data after clustering and the advertisement data after clustering; establishing a tensor and using a Tucker tensor decomposition method to decompose the tensor to obtain an approximate tensor after tensor dimension reduction; and performing supporting vector machine learning based on a radial basis function according to other target attributive characteristic data and the approximate tensor, and obtaining the advertisement click-through rate pre-estimation model. According to the invention, the relation between the users, the query key words and the advertisements are considered sufficiently, the mode combining the characteristic dimension reduction with the characteristic learning is used, and the advertisement click-through rate can be pre-estimated accurately.

Description

technical field [0001] The invention relates to the field of information prediction, in particular to a method and system for obtaining an advertisement click rate prediction model. Background technique [0002] Online advertising is one of the business models of the Internet. With the development of online promotion technology, advertising is changing from "extensive" delivery to "precision" delivery. The research in this field is called "computational advertising". Advertisement click-through rate (CTR, Click-Through-Rate) is an important index to evaluate the user's preference for advertisements, and it is of great significance for the precise placement of advertisements. Therefore, the click-through rate estimation is a hot issue in the fields of information retrieval and machine learning. . [0003] Search ads are characterized by a large amount of data and high-dimensional and sparse features. The dimension of effective information (non-zero value) in high-dimensional...

AI Technical Summary

Problems solved by technology

However, these methods do not take into account the internal relationship between the three-dimensional space of the user, the query and the advertisement presented by the advertisement click data while reducing the dimensionality of the data, which destroys the internal relationship between the three, causes a large loss of information, and cannot accurately predict the advertisement click rate. estimate

Images