A training method and device for a recommendation model, a recommendation method and device

Abstract

The present application provides a method and device for training a recommendation model, and a method and device for recommendation, wherein the method for training a recommendation model includes determining a first initial parameter value of a training parameter in the recommendation model to be trained, wherein the The first initial parameter value is the target parameter value after the initial parameter value is iteratively updated by the pre-trained click-through rate estimation model; the user characteristics of at least two sample users and the attribute characteristics of at least two sample applications are obtained; based on user characteristics, The attribute feature generates positive samples purchased by the sample user for the exposed sample application and negative samples that the sample user did not purchase from the exposed sample application; based on a sample set including at least one positive sample and a negative sample and the first initial parameter value The recommendation model to be trained is trained to obtain the recommendation model, and the recommendation model outputs the exposure conversion rate of each sample user to each sample application.

Description

technical field [0001] The present application relates to the field of computer technology, and in particular to a method and device for training a recommendation model, a method and device for recommendation, a computing device, and a computer-readable storage medium. Background technique [0002] In some Internet products, in order to facilitate users to quickly find the products they want (such as APP), personalized recommendation is required, and the APP that the user is most likely to click to buy is recommended to the user to be recommended. [0003] At present, some platforms simply use traditional machine learning models (such as LR or GBDT models) to make recommendations based on the tags of users and apps when making APP recommendations. This method of using tags to recommend requires a lot of manpower to establish Labeling system, and it is easy to recommend inappropriate apps to users (for example: apps for pornography, gambling and drugs), and it does not make f...

AI Technical Summary

Problems solved by technology

[0003] At present, some platforms simply use traditional machine learning models (such as LR or GBDT models) to make recommendations based on the tags of users and apps when making APP recommendations. This method of using tags to recommend requires a lot of manpower to establish Labeling system, and it is easy to recommend unsuitable apps to users (for example: apps for pornography, gambling and drugs), and it does not make full use of the purchase data after clicking, which may cause apps with high click-through rates to have low purchase conversion rates after users click

Images