Recommendation algorithm for relieving echo room problem based on causal inference

Through the methods of causal inference and counterfactual reasoning, pseudo-correlation and deviation are eliminated, and the mutual influence mechanism between variables in the recommendation system is clarified, which solves the explanatory and efficiency problems of the existing recommendation system when alleviating echo chamber problems, and achieves a more effective recommendation effect.

CN119719477AActive Publication Date: 2025-03-28TIANJIN UNIV
4 Cites 0 Cited by

Patent Information

Application Number
CN202411488002.X
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-10-24
Publication Date
2025-03-28
Estimated Expiration
2044-10-24

AI Technical Summary

Technical Problem

When existing recommendation systems alleviate echo chamber problems, they cannot effectively explain the mutual influence between variables and the mechanism of results changes, and are easily disturbed by pseudo-correlation, resulting in the impact of recommendation performance.

Method used

The recommendation algorithm based on causal inference is adopted to eliminate pseudo-correlation through causal graph analysis, and the backdoor adjustment of counterfactual reasoning is used to avoid project exposure deviation and user selection deviation caused by historical imbalance distribution.

Benefits of technology

The mutual influence mechanism between variables in the recommendation system is clarified, and a more interpretable and effective recommendation scheme is designed, which effectively alleviates the echo chamber problem and reduces the impact of pseudo-correlation on recommendation performance.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The invention relates to a recommendation algorithm for relieving an echo room problem based on causal inference, which eliminates pseudo-correlation among user characterization, item characterization and a model prediction result, and eliminates a wrong causal relationship caused by deviation. According to the method, firstly, analysis is carried out through a causal graph to obtain that false correlation is caused by project exposure deviation caused by historical unbalanced distribution, and secondly, a back door of anti-fact reasoning is adopted to adjust a depolarization mode to avoid the influence caused by the exposure deviation. According to the method, from the angle of causal relationship, mutual influence between variables and a result change mechanism are cleared, real reasons causing the echo room problem are explored, confusion factors among user characterization, project characterization and prediction results are removed, and the influence of the echo room problem on a recommendation system is further reduced.
Need to check novelty before this filing date? Find Prior Art