Model training method for removing position bias, medium, device and computing equipment

Abstract

The embodiment of the invention provides a model training method for removing position offset, a medium, a device and computing equipment, and the method comprises the steps: obtaining a quasi-predicted value corresponding to any task in a click task or a complete broadcast task in a model training process based on a main model according to user features and object features of a sample object in the task; meanwhile, based on the first slave model, obtaining a position influence value corresponding to the sample object through an object position feature of the sample object; and performing iterative training on the master model and the slave model through the obtained position influence value, so that the trained master model can remove the position offset, thereby eliminating the influence of the position of the object on a recommendation result when object recommendation is performed through the master model, obtaining a potential object which is really interested by a user, and improving the user experience. Therefore, the satisfaction degree of the user is improved, and the recommendation conversion rate of the platform is improved.

Description

technical field [0001] Embodiments of the present disclosure relate to the field of information processing technologies, and more specifically, embodiments of the present disclosure relate to a model training method, medium, device, and computing device for removing position bias. Background technique [0002] This section is intended to provide a background or context for implementations of the disclosure that are recited in the claims. The descriptions herein are not admitted to be prior art by inclusion in this section. [0003] With the continuous development of Internet products, in order to provide users with personalized recommendation services and achieve a personalized experience of "thousands of people, thousands of faces", recommendation systems have emerged as the times require. [0004] The core of the recommendation system is to estimate the conversion probability of the recommended object, and recommend the object that is considered to have a higher conversio...

AI Technical Summary

Problems solved by technology

[0005] However, factors such as the position of the object on the recommendation page will affect the user's choice, and the current recommendation system usually recommends according to the user's preference, which makes it difficult for the object recommended by the platform to meet the user's needs

Images