Self-normalized double robust estimation for debiasing position bias

Zhiyuan He; Weisheng Lin; Erlang Long; Qijia Li

doi:10.1117/12.2671824

21 March 2023 Self-normalized double robust estimation for debiasing position bias

Zhiyuan He, Weisheng Lin, Erlang Long, Qijia Li

Proceedings Volume 12609, International Conference on Computer Application and Information Security (ICCAIS 2022); 1260907 (2023) https://doi.org/10.1117/12.2671824
Event: International Conference on Computer Application and Information Security (ICCAIS 2022), 2022, ONLINE, ONLINE

Abstract

Position bias: The same item in different positions affects the user’s preferences, especially for items that are usually more difficult to see in the backward position. Standard double robust (DR) estimator in recommendation systems is widely used due to its accuracy and robust properties. However, the conventional DR estimator was doubtful of its highperformance variance. In this paper, we examined how to reduce the variance of the current DR estimator. Our main goal is to optimize the existing double robust estimation by the self-normalization method to reduce the variation. Our mathematical deduction shows that the self-normalized double robust estimation (SNDR) has less variance. The subsequent experimental steps demonstrate the low variance property of SNDR with both figure demonstration and data analysis.

Citation Download Citation

Zhiyuan He, Weisheng Lin, Erlang Long, and Qijia Li "Self-normalized double robust estimation for debiasing position bias", Proc. SPIE 12609, International Conference on Computer Application and Information Security (ICCAIS 2022), 1260907 (21 March 2023); https://doi.org/10.1117/12.2671824

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available